eWEEK https://www.eweek.com/ Technology News, Tech Product Reviews, Research and Enterprise Analysis Fri, 05 Jan 2024 21:45:03 +0000 en-US hourly 1 https://wordpress.org/?v=6.3.2 Generative AI Models: A Complete Guide https://www.eweek.com/artificial-intelligence/generative-ai-model/ Fri, 05 Jan 2024 14:00:45 +0000 https://www.eweek.com/?p=222621 Generative AI models are powerful tools for producing data. Learn what generative AI models are, how they work & what benefits they offer.

The post Generative AI Models: A Complete Guide appeared first on eWEEK.

Generative artificial intelligence (AI) models are AI platforms that generate a variety of outputs based on massive training datasets, neural networks, deep learning architecture, and prompts from users.

Depending on the type of generative AI model you’re working with, you can generate images, translate text into image outputs, synthesize speech and audio, create original video content, and generate synthetic data.

While many AI companies and tools are popping up daily, the generative AI models that work in the background to run these tools are fewer and more important to the growth of generative AI’s capabilities. Todays’ generative AI models are the “unsung” heroes of AI.

Read on to learn more about generative AI models, how they work and compare to other types of AI, and some of the top generative AI models available today.

How Do Generative AI Models Work?

Generative AI models are the massive, big-data-driven models that power the emerging artificial intelligence technology that can create content.

Using unsupervised or semi-supervised learning methods, generative AI models are trained to recognize small-scale and overarching patterns and relationships in training datasets that come from all kinds of sources — the internet, wikis, books, image libraries, and more.

This training enables a generative AI model to mimic those patterns when generating new content, making it believable that the content could have been created by or belonged to a human rather than a machine.

Generative AI models are able to so closely replicate actual human content because they are designed with layers of neural networks that emulate the synapses between neurons in a human brain. When the neural network design is combined with large training datasets, complex deep learning and training algorithms, and frequent re-training and updates, these models are able to improve and “learn” over time and at scale.

Among the many types of generative AI models, there are text-to-text generators, text-to-image generators, image-to-image generators, and even image-to-text generators. In this example, I used a text-to-image generator, Img2Go. I provided an AI prompt – a text description – and the model generated a new image that matched my prompt.

The prompt I used was: “A laughing robot in the sunset.”

AI-generated image of a laughing robot in the sunset.
An example of the output of a generative AI model, prompted with the phrase, “A laughing robot in the sunset.”

For more information about how generative AI is used in business, see our guide: Generative AI Examples

How Are Generative AI Models Trained?

Generative AI models are all trained a little differently, depending on the model type you’re training. Let’s look at how transformer-based models, GANs, and diffusion models are trained:

Transformer-Based Model Training

Transformer-based models are designed with massive neural networks and transformer infrastructure that make it possible for the model to recognize and remember relationships and patterns in sequential data.

To start, these models are trained to look through, store, and “remember” large datasets from a variety of sources and, sometimes, in a variety of formats. Training data sources could be websites and online texts, news articles, wikis, books, image and video collections, and other large bodies of data that provide valuable information.

From there, transformer models can contextualize all of this data and effectively focus on the most important parts of the training dataset through that learned context. The sequences this type of model recognizes from its training will inform how it responds to user prompts and questions.

Essentially, transformer-based models pick the next most logical piece of data to generate in a sequence of data. Encoders and/or decoders are built into the platform to decode the tokens or blocks of content that have been segmented based on user inputs.

A general framework for Transformer-based language model pre-training.
A general framework for Transformer-based language model pre-training.

GAN Model Training

GAN (generative adversarial network) models are trained with two different sub-model neural networks: a generator and a discriminator. The generator generates content based on user inputs and training data, while the discriminator model evaluates generated content against “real” examples to determine which output is real or accurate.

First, the generator creates new “fake” data based on a randomized noise signal. Then, the discriminator blindly compares that fake data to real data from the model’s training data to determine which data is “real” or the original data.

The two sub-models cycle through this process repeatedly until the discriminator is no longer able to find flaws or differences in the newly generated data compared to the training data.

General structure of the GAN training process.
General structure of the GAN training process.

Diffusion Model Training

Diffusion models require both forward training and reverse training, or forward diffusion and reverse diffusion.

The forward diffusion process involves adding randomized noise to training data. The model is trained to generate outputs using the noisy data (not as refined or specific) as input. The noise introduces variations and perturbations in the data, making the model robust and helping it to learn different possible outputs for a given input.

When the reverse diffusion process begins, noise is slowly removed or reversed from the dataset to generate content that matches the original’s qualities. This process encourages the model to focus on the underlying structure and patterns in the data, rather than relying on the noise to produce the desired outputs. By gradually removing the noise, the model learns to produce outputs that closely match the desired qualities of the original input data.

Noise, in this case, is best defined as signals that cause behaviors you don’t want to keep in your final dataset but that help you to gradually distinguish between correct and incorrect data inputs and outputs.

For a related AI chatbot comparison, see: ChatGPT vs. GitHub Copilot

Types of Generative AI Models

Many types of generative AI models are in operation today, and the number continues to grow as AI experts experiment with existing models.

With the classifications below, keep in mind that it’s possible for a model to fit into multiple categories. For example, the latest updates to ChatGPT and GPT-4 make it a transformer-based model, a large language model, and a multimodal model.

  • Generative adversarial networks (GANs): Best for image duplication and synthetic data generation.
  • Transformer-based models: Best for text generation and content/code completion. Common subsets of transformer-based models include generative pre-trained transformer (GPT) and bidirectional encoder representations from transformers (BERT) models.
  • Diffusion models: Best for image generation and video/image synthesis.
  • Variational autoencoders (VAEs): Best for image, audio, and video content creation, especially when synthetic data needs to be photorealistic; designed with an encoder-decoder infrastructure.
  • Unimodal models: Models that are set up to accept only one data input format; most generative AI models today are unimodal models.
  • Multimodal models: Designed to accept multiple types of inputs and prompts when generating outputs; for example, GPT-4 can accept both text and images as inputs.
  • Large language models: The most popular and well-known type of generative AI model right now, large language models (LLMs) are designed to generate and complete written content at scale.
  • Neural radiance fields (NeRFs): Emerging neural network technology that can be used to generate 3D imagery based on 2D image inputs.

Generative AI vs. Discriminative AI Models

The primary difference between generative and discriminative AI models is that generative AI models can create new content and outputs based on their training.

Discriminative modeling, on the other hand, is primarily used to classify existing data through supervised learning. As an example, a protein classification tool would operate on a discriminative model, while a protein generator would run on a generative AI model.

Generative vs. Predictive AI Models

Generative models are designed to create something new while predictive AI models are set up to make predictions based on data that already exists.

Continuing with our example above, a tool that predicts the next segment of amino acids in a protein molecule would work through a predictive AI model while a protein generator requires a generative AI model approach.

What are the Challenges of Generative AI Models?

Even though generative AI has been trending since November 2022, one of the major reasons you see only a limited number of startups developing AI models is that it requires deep pockets and vast resources and is very complex. We highlighted some key challenges of generative AI models below.

Mode Collapse in GANs

GANs may suffer from mode collapse, which is when the generator learns to fool the discriminator by producing a limited set of outputs, ignoring the diversity present in the training data. This can result in repetitive or less varied generated content.

Training Complexity 

As stated above, generative models often require large amounts of data and computational resources for training. The resource-intensive nature of training limits accessibility for smaller research labs and individual researchers. It also requires domain-specific knowledge, as a lack of domain-specific expertise can result in the AI model giving suboptimal outputs, or even hallucinating.

Adversarial Attacks

Generative models, especially GANs, are susceptible to adversarial attacks where small perturbations to input data can lead to unexpected or malicious outputs. Learning to effectively combat adversarial attacks is an active area of research.

Fine-tuning and transfer learning

Adapting pre-trained generative models to specific tasks or domains may be challenging. The ability to fine-tune without causing catastrophic forgetting or degradation in performance is an another ongoing research concern, and requires more work and investment.

To learn about the companies actively developing generative AI, see our guide: Generative AI Companies: Top 12 Leaders

What are the Benefits of Generative AI Models?

The benefits of generative AI models are numerous and very important to AI’s growth, particularly in the area of data augmentation and natural language processing.

Data Augmentation

Generative models can be used to augment datasets by generating synthetic data. This is helpful in scenarios where sufficient real-world labeled data is not available, making it useful for training other machine learning models.

Natural Language Understanding and Generation

Generative AI models can be used to create AI chatbots and virtual AI assistants capable of understanding and generating human-like responses in natural language. It can generate human-like text for content creation, including articles, stories, and more.

Creative Applications

Generative AI models can be used to create art, poetry, music and other artistic works. For example, OpenAI’s Jukebox, a generative model, can compose music in different genres. Generative AI models can also be leveraged for content synthesis, as it is capable of producing diverse and creative content, and assisting in brainstorming and ideation processes.


AI models can be fine-tuned for various tasks, such as translation, summarization, and question answering. They can also be adapted to different domains and industries with proper training and fine-tuning.

For example, depending on the tuning, the output very serious and proper, or casual and recreational. The mode and mood of the output can be tuned to a remarkably specific degree.

Examples of Generative AI Models

Below you’ll find some of the most popular generative AI models available today. Keep in mind that many generative AI vendors build their popular tools with one of these models as the foundation or base model. For example, many of Microsoft’s new Copilot tools run on GPT-4 from OpenAI.

  • GPT-3/3.5/4, etc.: GPT-3, GPT-3.5, and GPT-4 are different generations of the GPT foundation model managed, owned, and created by OpenAI. The latest version, GPT4, uses a multimodal LLM that is the basis for ChatGPT.
  • OpenAI Codex: Another model from OpenAI, Codex is able to generate code and autocomplete code in response to natural language prompts. It is the foundation model for tools like GitHub Copilot.
  • Stable Diffusion: One of the most popular diffusion models, Stability AI’s Stable Diffusion is primarily used for text-to-image generation.
  • LaMDA: A transformer-based model from Google, LaMDA is designed to support conversational use cases.
  • PaLM: Another transformer-based LLM from Google, PaLM is designed to support multilingual content generation and coding. PaLM 2 is the latest version of the model and is the foundation for Google Bard.
  • AlphaCode: A developer and coding support tool from DeepMind, AlphaCode is a large language model that generates code based on natural language inputs and questions.
  • BLOOM: Hugging Face’s BLOOM is an autoregressive, multilingual LLM that mostly focuses on completing statements with missing text or strings of code with missing code.
  • LLaMA: LLaMA is a smaller large language model option from Meta, with a goal of making generative AI models more accessible to users with fewer infrastructural resources.
  • Midjourney: Midjourney is a generative AI model that operates similarly to Stable Diffusion, generating imagery from natural language prompts that users submit.

Keep learning about AI: Generative AI Landscape: Current and Future Trends

What Can Generative AI Models Do?

Generative AI models support various use cases, allowing you to complete a variety of business and personal tasks when trained appropriately and given relevant prompts. You can use generative AI models to handle the following tasks and many more:

Language Visual Auditory
Generate and complete text Image generation Music generation
Code documentation Video generation Voice synthesis
Answer questions and support research 3D models Voice cloning
Design proteins and drug descriptions Optimize imagery for healthcare diagnostics
Supplement customer support experiences Create immersive storytelling and video game experiences
Generate synthetic data Synthetic media
Code generation Procedural generation

Bottom Line: Generative AI Models

Generative AI models are highly scalable, accessible artificial intelligence solutions that are rightfully getting publicity as they supplement and transform various business operations.

However, there are many concerns about how these tools work, their lack of transparency and built-in security safeguards, and generative AI ethics in general. Whether your organization is working to develop a generative AI model, build off of a foundational model, or simply use ChatGPT for daily tasks, keep in mind that the best way to use generative AI models is with comprehensive employee and customer training and clear ethical use policies in place.

For a deeper understanding of the AI chatbot sector, see our guide: Best AI Chatbots 2024

The post Generative AI Models: A Complete Guide appeared first on eWEEK.

ChatGPT-4 vs. Claude: 2024 Chatbot Comparison https://www.eweek.com/artificial-intelligence/gpt4-vs-claude/ Fri, 05 Jan 2024 13:37:32 +0000 https://www.eweek.com/?p=222309 Compare the features of Claude and ChatGPT in this comprehensive comparison. Get up-to-date information on which chatbot best fits your needs.

The post ChatGPT-4 vs. Claude: 2024 Chatbot Comparison appeared first on eWEEK.

ChatGPT-4 and Claude are both leading AI applications that use generative AI. GPT-4 is the latest version of ChatGPT, the popular AI application that can create articles and computer code and perform various tasks. GPT-4 is supported by OpenAI and Microsoft.

Claude is created by Anthropic and is a newer application in the AI field. It’s used mainly as an AI chatbot, and doesn’t have the full functionality of GPT-4. Claude is supported by Google, Zoom and Slack, and has received significant investment over the last several months.

So which generative AI app is best, GPT-4 or Claude 2? In this comparison of ChatGPT and Claude, we analyze both application’s key features, pricing, and pros and cons to help you determine the best AI solution for your needs.

GPT-4 vs. Claude: Head-to-Head Comparison

Here is a head-to-head feature comparison of GPT-4 and Claude.

GPT-4 Claude
Developed by OpenAI Anthropic
Pricing $20 a month, plus additional fees for volume used $20 per month plus tax
Free plan Yes with ChatGPT, No with GPT-4 Yes (Beta model)
Language availability 200+ 10+
Long-form content Better Good
Maximum context length 32k (32,000) tokens 100,000 tokens
Chatbot functions Good Good
Data processing Superior Adequate
Visual input Yes No
Originality rate High Moderate
Image Interpretation Good Absent
Parameters analyzed Trillions from the web Limited to specific data sets requested
Ethics N/A Constitution-inspired responses
Integration Good Very good
Programming accuracy Good Fair but basic
Use cases ChatGPT, marketing content, Copilot Legal analysis, Math, content moderation

OpenAI icon.

What is GPT-4?

ChatGPT is an abbreviation for Generative Pre-trained Transformer, a form of advanced artificial intelligence (AI). GPT4 simulates thought by using a neural network machine learning model trained on a vast trove of data gathered from the internet.

GPT-4 goes far beyond being a chatbot. This generative AI app creates documents and articles and solves problems. GPT-4 can also do image interpretation using multimodal language AI models. This enables it to build websites based on sketches and suggest recipes based on a photo of what is in the fridge or sitting on a countertop.

Below is an example that demonstrates that GPT-4 provides more detailed responses than its earlier version:

GPT-4 provides more detailed answers than GPT-3

For more information, also see: ChatGPT: Understanding the ChatGPT ChatBot

Anthropic icon.

What is Claude 2?

Claude 2 is an AI assistant from Anthropic that is accessible through a chat interface or API. It is capable of conversational and text processing. Use cases include summarization, search, creative and collaborative writing, formulating Q&A and some basic coding.

It gives fast responses to customer service requests and can hand off tasks to a human when needed. Claude is particularly good at editing, rewriting, summarizing, classifying and extracting structured data. It can also follow basic instructions and logical scenarios, analyze strategic risks and opportunities based on annual reports, assess the pros and cons of a piece of legislation and identify risks in legal documents.

Claude can provide lengthy answers to questions requiring fact or opinion. Below is Claude’s answer to “what is the best weather?”

An answer requiring an opinion, provided by Claude.

GPT-4 vs. Claude 2: Feature Comparison

GPT4 and Claude are at the center of the generative AI battle royale in the technology sector. Let us compare their key features to help determine the best option for you.

GPT-4 vs. Claude: Chatbot Functions

GPT-4 is used a lot in chatbot applications to automate customer service, answer FAQs, and engage in conversation. It can respond conversationally by tapping into a comprehensive set of online text, as well as news items, novels, websites, and more.

Overall, GPT4 does a good job of analyzing information, evaluating online behavior, and even makes product recommendations as part of the online sales and upselling process. Automation features extend to appointment scheduling, reservations, payment processing, queries about shipping schedules, order progress, product returns, product and service availability, all with a good level of accuracy.

Claude can be viewed as more of a focused AI chatbot than GPT-4. It can be tailored to certain use cases and data sets such as customer service, legal, back office and sales. It can also be taught when to hand off tasks to a human.

You define the data set to analyze, summarize or use for context, and it can respond conversationally. It can speak a variety of common languages, as well as programming languages. However, it doesn’t search the internet like GPT-4. You can provide Claude with text from the internet and ask it to perform tasks with that content.

Verdict: GPT-4 wins overall – but Claude may prove better at narrower AI chatbot use cases.

GPT-4 vs. Claude: Accuracy of Response

ChatGPT can be prone to be error-based, because some of its data may not be current. But most of the time it is accurate. GPT-4 added a greater degree of accuracy. OpenAI stated that GPT-4 is 82% less likely than its predecessor to respond to requests for content that OpenAI does not allow, and 60% less likely to invent answers. But don’t expect it to be perfect. That includes coding – its programming output should always be verified by human eyes.

Claude is also prone to errors. Some believe Claude is better at answering queries as it takes answers from a narrower data set. This is beneficial in areas such as history, geography and entertainment. A big difference between Claude and GPT-4 is that Claude can admit that it doesn’t know.

Verdict: This category is even. GPT4 is more accurate than Claude in many areas, yet Claude’s accuracy is better in some use cases.

On a related topic: What is Generative AI?

GPT-4 vs. Claude: Integration

GPT can be plugged into other applications to generate responses via an API. Plugins are becoming available, including those for the likes of Kayak, Expedia, OpenTable, Slack and Shopify, with more on the way. It is also integrated with a lot of different programming languages.

Claude has done an excellent job of developing partners, too. They include Notion, Quora, DuckDuckGo, Slack and Zoom. Watch out for tighter Google integration going forward, too. Claude can integrate with a wide range of apps via API.

The Claude App for Slack can summarize threads and answer questions. For Zoom, Claude is being integrated into the Zoom contact center, meetings, phone, team chat, whiteboard, and Zoom IQ.

Verdict: Both integrate well, but Claude perhaps comes out slightly ahead.

For more information about the overall AI software sector: Top AI Software

GPT-4 vs. Claude: Security

Unfortunately, there’s not a lot to say about the security of these two AI platforms, because they are so new, and security has not been a driving focus as they’ve been built. Expect this to change going forward, but in any case, it’s never a good idea to create AI prompts from sensitive business information.

Chat-GPT does claim that it has worked with over 50 experts to develop improved security.  Claude, to its credit, uses industry-standard best practices for data handling and retention. But, again, neither AI platform is intended for private business information.

Verdict: Neither is particularly well known for security, but Claude gets the nod.

For more comparisons of generative AI applications: ChatGPT vs. Google Bard

GPT-4 vs. Claude: Pricing

GPT-4 pricing

GPT-4 has a basic version available for free but the main ChatGPT Plus version costs roughly $20/month. Subscribers gain access to ChatGPT at peak times, faster responses, and priority access to new features and improvements.

On top of the basic subscription, there is a pricing scale per 1,000 tokens (chunks of words). 1,000 tokens comes out to about 750 words of material. The higher rate provides access to a larger set of contextual data.

Model Input Output
GPT-4 $0.03 per 1K tokens $0.06 per 1K tokens
GPT-4-32k $0.06 per 1K tokens $0.12 per 1K tokens

Claude Pricing

Claude 2 has a subscription and a pricing scale options. The subscription plan, Claude Pro, costs $20 per month plus tax. There are three scale versions of Claude. Claude Instant is a lighter, less expensive and much faster option. Claude 2 is the high-performance version that can take care of sophisticated dialogue and creative content generation as well as detailed instruction following. Claude 2.1 offers the same performance as Claude 2 with reduction in hallucination rates.

For context, 100,000 tokens works out to roughly 70,000 words or six hours of audio.

Model Context window Input (Prompt) Output (Completion)
Claude Instant 100,000 tokens $0.80 per million tokens $2.40 per million tokens
Claude 2 100,000 tokens $8.00 per million tokens $24.00 per million tokens
Claude 2.1 200,000 tokens $8.00 per million tokens $24.00 per million tokens

Due to the added monthly subscription in GPT-4, it is hard to be certain who wins in the long term. But it looks like Claude 2 wins on pricing.

Pros and Cons of GPT-4

ChatGPT has a higher profile than Claude – more total users – yet it is also prone to hallucinations, and is a little slower.

Pros Cons
GPT-4 can process different text files like PDFs, CSVs, TXT, Markdown or images, audio More likely to provide nonsensical statements
Supports 200+ languages Slow response time compared to Claude
Public to all customers who have OpenAI account
Wider general knowledge

Pros and Cons of Claude 2

Claude is known to be a fast AI chatbot, but its list of languages is far smaller.

Pros Cons
Fast response to questions Limited to 10 languages
Provide step-by-step and logical explanation to mathematical questions Limited availability
Superior performance in fields like law, mathematics, and coding
Avoiding contradictory or nonsensical statements

GPT-4 vs. Claude 2: Which is the Best?

The decision of whether to choose GPT-4 or Claude 2 depends on your needs, priorities and budget. To break it down, Claude 2’s standout feature is its ethical framework, while GPT-4 excels as a generalist.

GPT4 can perform complex tasks. It has achieved some success with basic computer programming duties. It ventures well beyond that into territory such as drawing up simple lawsuits, creating elementary computer games, passing exams, checking for plagiarism, generating written content, summarizing documentation, highlighting key passages within the text, and translating languages.

On the other hand, Claude 2 shines in key areas like legal applications, math and science, and safety. Its constitutional AI enables it to understand and apply ethical principles and guidelines when providing output. This makes Claude 2 a suitable choice for professionals in fields that require adherence to ethical standards, such as legal practitioners or researchers in sensitive scientific areas.

For comprehensive conversational applications, GPT-4 still leads in general performance whereas Claude 2’s strengths make it a worthy challenger that can go toe-to-toe with GPT4 in various disciplines.

For a fuller understanding of the generative AI market: Top Generative AI Apps and Tools

Bottom Line: GPT-4 vs. Claude 2

GPT-4 uses a transformer-based architecture as part of a neural network that handles sequential data. It appears to perform capably in coding, and does a very good job on chat, language translation, answering questions, understanding images, and can even determine why a joke is funny. GPT-4 has thrown its hat in the ring with Microsoft – which offers many benefits in terms of development and support.

Claude, on the other hand, favors Google. It is less comprehensive than GPT-4 but can respond without searching the internet. It does a good job of digesting, summarizing and explaining financial documents and research papers. Like all AI chatbots, it may sometimes assess its own ability or memory incorrectly, make up information, and make errors in complicated arithmetic. But certainly GPT-4 makes errors, too.

Claude is very much a work in progress, it’s not as comprehensive as GPT-4 right now. Give it a few months and we’ll see how far it comes. But for now, GPT-4 has greater functionality.

For more about new companies that are influencing the future of AI: Generative AI Startups

The post ChatGPT-4 vs. Claude: 2024 Chatbot Comparison appeared first on eWEEK.

How FinOps and AI Curb Escalating Cloud Costs https://www.eweek.com/artificial-intelligence/how-finops-and-ai-curb-escalating-cloud-costs/ Thu, 04 Jan 2024 19:21:00 +0000 https://www.eweek.com/?p=223614 A new report from Tangoe sheds light on FinOps implementations, artificial intelligence, and how to improve cloud costs and financial predictability. Cloud has taken every industry by storm. But the report shows that, despite the apparent benefits, “a variety of challenges still get in their way, mostly around cost control, security expertise, and a skills […]

The post How FinOps and AI Curb Escalating Cloud Costs appeared first on eWEEK.

A new report from Tangoe sheds light on FinOps implementations, artificial intelligence, and how to improve cloud costs and financial predictability.

Cloud has taken every industry by storm. But the report shows that, despite the apparent benefits, “a variety of challenges still get in their way, mostly around cost control, security expertise, and a skills gap.” Furthermore, the ongoing “cloud sprawl” has not helped these issues. In fact, they’ve gotten worse. There’s a lack of visibility, accountability, and governance of the costs of SaaS and IaaS.

The report, based on a survey by Foundry of 200 enterprise IT decision-makers in a number of industries, looked at the benefits of FinOps, which is a framework that maximizes the business value of cloud computing. 

Overcoming Cloud Challenges with FinOps

FinOps can be a path to overcoming cloud cost challenges, with respondents reporting these are the main drivers for using FinOps:

  • 70% say they’re interested in increasing cloud resource performance.
  • 60% say they want to make budget cuts.
  • 58% say they want to manage rising costs and macroeconomic pressures.

When it comes to the benefits, the survey showed increased productivity (44%), cost savings (43%), and reduced security risk (43%) at the top of the list.

Implementations vary widely, but there are three factors, according to Foundry, that increase the likelihood of success: implementation approach, the use of artificial intelligence, and wide-reaching optimization programs spanning both IaaS and SaaS.

For a deeper portrait of the cloud market, read our in-depth overview: Top Cloud Service Providers and Companies

In-House vs. Outsourced

When it comes to in-house vs. outsourced implementations, outsourced is the clear winner, with in-house FinOps solutions fetching less than 10% in savings and outsourced solutions bringing in 20% savings.

Tangoe’s chief product officer, Chris Ortbals, quoted in the report, says that although DIY is the way most companies go, they’re not best suited to optimize those implementations. “Strategic partners are essential because they look at usage and expenses from a different perspective,” he said. This diverse vantage point provides them with “the context of understanding how hundreds of companies optimize millions of dollars in IT spending.”

AI in FinOps

With AI top-of-mind for everyone these days, this report doesn’t disappoint. The survey shows that 71% of respondents use or plan to use AI in their programs.

That makes sense when you see that companies that use AI in FinOps are 53% more likely to save more than 20%. On the other side of the ledger, companies that don’t use AI will get savings of only 6% to 10%.

Ortbals adds that every FinOps practitioner should include AI in their bag of tricks. “Without AI it’s simply impossible to evaluate massive amounts of data against all possible configuration options, playing out what-if scenarios to quickly determine which action will yield the highest savings,” he said.

“AI can do this at enterprise scale. Plus, it can help IT engineers implement cost-saving recommendations once approved. This results in faster time-to-savings and programs that can achieve higher returns.”

The report says 63% of respondents think analytics is the top use case for AI in FinOps, with 50% saying it eases the FinOps management burden and 48% saying they have seen productivity gains from AI automation.

Also see: 100+ Top AI Companies 2024

SaaS and IaaS

When it comes to SaaS and IaaS, the report shows SaaS users can save 20% or more while IaaS users save less than 10%.

Ortbals said that savings add up with one high-volume application or a few smaller applications that can break a budget.

“But CIOs and CFOs should drive broader cloud ROI,” he said. “FinOps is designed to maximize the benefits of SaaS and IaaS. Start with applications, expand into cloud infrastructure, and consider how you can apply FinOps principles to other areas of IT spending like mobile and telecom.”

Choosing a FinOps Solution

So, what should you consider when selecting a solution? The report says that these are the top five criteria:

  • 70%: Industry expertise. Customers want a partner who is versed in the complexities of cloud optimization but also understands the specific insights of each vertical, as mapping them to FinOps is mandatory for success.
  • 69%: AI and automation capabilities. There is no bigger transformative technology today than AI, which will have a massive impact on FinOps. Basic use cases include automation of basic tasks, but long-term, generative AI will enable users to access information via natural language.
  • 68%: Fully managed services. Not all businesses have the desire to operate FinOps themselves, particularly with AI accelerating the pace of change. Managed services are an excellent option for companies that want to take advantage of FinOps but do not want to go through the deployment and management process.
  • 63%: Flexibility of solution. Businesses want choices, and vendors that offer optimization for only one or two cloud service providers or a limited number of applications are quickly outgrown as customers look to cut costs across their rapidly changing multicloud estates.
  • 62%: Certifications, licenses, awards. These validate that the vendor solutions work as advertised and have the leading features to give customers a competitive advantage.

Bottom Line: The Value of FinOps

This report provides some fascinating insights for cloud users. FinOps is a great use case for AI, and managing cloud resource utilization could be one of the more intriguing implementations of the technology, with significant savings for enterprises possible in the near term.

To learn more about how companies are modernizing their IT processes, read our in-depth article: Digital Transformation Guide

The post How FinOps and AI Curb Escalating Cloud Costs appeared first on eWEEK.

Chronosphere’s Ian Smith on Cloud Native Observability https://www.eweek.com/cloud/chronospheres-ian-smith-on-cloud-native-observability/ Wed, 03 Jan 2024 23:44:38 +0000 https://www.eweek.com/?p=223579 I spoke with Ian Smith, Field CTO at Chronosphere, about cloud native observability, an emerging tech that’s gaining a lot of attention in the enterprise. Smith spoke in-depth about trends and best practices in observability. An observability solution monitors a company’s IT infrastructure by constantly monitoring its outputs. In most cases, the most important outputs […]

The post Chronosphere’s Ian Smith on Cloud Native Observability appeared first on eWEEK.

I spoke with Ian Smith, Field CTO at Chronosphere, about cloud native observability, an emerging tech that’s gaining a lot of attention in the enterprise. Smith spoke in-depth about trends and best practices in observability.

An observability solution monitors a company’s IT infrastructure by constantly monitoring its outputs. In most cases, the most important outputs are those that track the performance of the core applications that enable the company to keep running.

Historically, this monitoring task has been handled by humans, and still is, largely. Yet as tech infrastructure grows ever more complex, humans need ever more help to keep up. Hence the growth of observability solutions.

In fact, observability is increasingly important as IT infrastructure grows increasingly complex – and companies know that.

Chronosphere focuses on cloud native observability, which is well-suited  to contemporary IT deployments that include elements like microservices and multicloud deployments.

See below for a podcast and video version of the interview.

What’s Driving Observability Adoption

 As recently as 2021, just 61 percent of companies had a centralized observability solution. By 2023, that number had risen to about 70 percent. So certainly observability has room for growth – and yet there’s still some skepticism about this emerging tech.

As Smith noted, the doubting companies ask, “‘well, what am I getting beyond the high-level marketing message?’”

The answer, ultimately, is that observability is far more than cobbling together an array of  interoperating tools. What drives companies to adopt this technology is how observability can assist engineering to facilitate what a business needs, and also – especially – observability’s ability to help control costs, including the expense of multicloud computing.

It’s also about growth. Smith has heard companies say, “’When we previously settled on our observability tooling, we were a much smaller company. We had a really big focus on observability, used by our most senior resources – and they drove the evaluation. Now we have maybe 10, 15 times more engineers and it’s a very broad spread of experiences.”

Observability Strategy

Everything in enterprise IT requires planning, but observability, due to its complexity, requires truly deliberate planning.

A company needs to understand, Smith noted, “Who’s using [the observability tool], what are they using it for, how much are they utilizing it? And be able to compare across those data sets.”

For instance, “Maybe you have some data that’s only used once every three months for a capacity plan, but it’s very, very small in its footprint. But there’s data over here, it’s used every day, for instance for [important] investigations.” It’s essential to know where your data is – and what data is needed at all time to answer essential business questions.

The most important element of creating an observability strategy is deciding precisely what your goal is – and agreeing on that goal across the company.

So ask, “What is the problem we’re actually trying to solve?” Smith said. Some companies aren’t all on the same page. “And so how can you possibly buy something thinking it’s a solution for a problem if you haven’t actually completely defined that [data] problem upfront?

“Maybe it is, for instance, that we need to direct a large portion of observability data into some other area, maybe a data lake because we’ve been abusing our observability tooling. And these are all strategic initiatives that come out of really stepping back and looking at that bigger picture.”

AI and the Future of Observability  

There is, in Smith’s view, a major industry hope that that AI can simply swoop down and answer all of the thorny issues involved with monitoring IT infrastructure. While that belief is unrealistic, certainly the growth of AI has major ramifications for IT – particularly due to AI’s assistance with communication between humans and the system.

In short, the future of observability will enable IT admins to simply talk to their systems. “Wouldn’t it be great to be able, in natural language, to really just ask what is going on with this particular part of the system? Then having a way for the observability system to distill down, ‘these are things you should be focusing on, and here’s an explanation for why.’”

This process is a realistic expectation for observability users. “It’s rooted in data and it’s rooted in building up [data] over time and understanding a model of what these things mean.”

Listen to the podcast:

Also available on Apple Podcasts

Watch the video:


The post Chronosphere’s Ian Smith on Cloud Native Observability appeared first on eWEEK.

How the PGA TOUR Uses Generative AI https://www.eweek.com/artificial-intelligence/how-the-pga-tour-seek-to-use-generative-ai/ Wed, 03 Jan 2024 21:14:55 +0000 https://www.eweek.com/?p=223610 Generative AI is a powerful tool for the PGA Tour. Learn how the PGA Tour is using generative AI to improve their operations and enhance the fan experience.

The post How the PGA TOUR Uses Generative AI appeared first on eWEEK.

At the recent AWS re:Invent conference in Las Vegas, AWS took a handful of press and analysts to the TPC Summerlin golf course in Las Vegas to understand how the PGA TOUR is considering using AI.

On the driving range, the PGA TOUR and AWS had set up a Generative AI demo with Scott Gutterman, Senior VP of Digital Operations with the PGA TOUR, and a handful of PGA TOUR and LPGA tour golf pros.

He presented a virtual interface built on data from the PGA TOUR’s ShotLink, powered by a CDW system, collected from 2005 to 2023. It included a massive 160,000 hours of video from their media asset management system, and 30 years of media guides in PDFs.

The application is built on various AWS tools, including Amazon Bedrock with the Anthropic Claude 2.0 Instant model, Amazon Kendra, and Amazon Athena.

Gutterman explained that the PGA TOUR and AWS have been working on creating a new experience with the AWS Generative AI Innovation Center for eight months. As impressive as the demo was, the potential is still greater. “We’ve all heard about ChatGPT and other platforms, and that’s caused many people to use these tools,” he said. “The PGA TOUR has been busy looking at what we can do and create for our players, fans, broadcasters, and for us internally.”

Generative AI: Interacting With the PGA Using Natural Language

The generative AI interface was demonstrated through a series of use cases where a question was posed, and the AI tool answered. Then the TOUR pro in attendance would try to replicate the golf shot in real time.

First up was the question, “What was Collin Morikawa’s longest drive at the Shriners Children’s Open?” which has been played at TPC Summerlin for several years. The bot answered that in 2019, Morikawa hit a 334-yard drive on the 15th hole. In addition to the information, the interface provided the actual TV footage from that shot. Morikawa then had three attempts to hit a drive that far.

Gutterman asked a series of questions, including Tony Finau’s longest drive at the Shriners Children’s Open, Charley Hull’s eagle, and Rose Zhang’s birdie from the 2023 US Women’s Open at Pebble Beach. In each case, the interface returned the relevant information and a video asset of the shot.

Generative AI Simplifies Search

After the demonstrations, I asked Gutterman about the generative aspect of what was shown. He explained that they’ve been able to search the data for some time, but you had to be precise about what you queried, and the response would be very specific.

In contrast, using large language models allowed the user to interact with natural language and then receive a response in natural language. Also, the tool packages up the relevant text and combines it with the video, providing more information to the user.

This is still in the beta phase, but Gutterman did talk about the tool’s potential. “The potential is massive,” he said. “When the players walk off the course, they could ask their personal bot how they did, and it could return a full performance report to help them prepare for the next round. They can learn much about what is happening within their game and what to work on.”

Gutterman also talked about the potential for the audience. “Our broadcasters could have better information on how each player performs on every course. They could use the data to help fans understand shot percentages, the likelihood of making putts, and more. For the fans, they could get information on players without digging through media guides and other data sources.”

Fans Gets an Enhanced Experience

Personally, I’m most excited about the fan-facing aspect of generative AI’s use by the PGA. When one watches a PGA TOUR event on TV, it’s hard, particularly for a casual fan, to fully understand how difficult the shots are. If, for instance, a viewer knew that a certain shot had only a 10% chance of success, that could add a lot of tension and excitement to the experience. AI enables the untrained eye to see things that experts can, and that’s a win for everyone.

Post-event, golf pro Adam Hadwin was kind enough to stay and chat with a few of us. We asked him about the potential. “The analytics can help us understand how to play certain shots,” he said. “As an example, the data might show that you’re better off missing the fairway left than right, so a player would then favor the left side. Or with a certain pin placement, the data might show that players three-put most often from a certain spot, so you’ll avoid that with your approach shot.”

Bottom Line: Generative AI is Changing Sports

The artificial intelligence era has arrived and will change almost every aspect of our lives. It was fascinating to see how the PGA TOUR is thinking of making its data more accessible to more audiences using generative AI.

It’s a good lesson for all businesses. Every company has massive amounts of data today, but it’s only useful if people can access the insights easily, and generative AI makes that possible.

The post How the PGA TOUR Uses Generative AI appeared first on eWEEK.

eWEEK TweetChat, January 16: Governing Generative AI https://www.eweek.com/artificial-intelligence/eweek-tweetchat-january-16-governing-generative-ai/ Tue, 26 Dec 2023 20:47:34 +0000 https://www.eweek.com/?p=223603 On Tuesday, January 16 at 11 AM PT, eWeek will host its monthly #eWEEKChat. The topic will be Governing Generative AI, and it will be moderated by James Maguire, eWEEK’s Editor-in-Chief. We’ll discuss – using X, formerly known as Twitter – challenges, issues, and best practices for governing artificial intelligence, which offers vast potential but […]

The post eWEEK TweetChat, January 16: Governing Generative AI appeared first on eWEEK.

On Tuesday, January 16 at 11 AM PT, eWeek will host its monthly #eWEEKChat. The topic will be Governing Generative AI, and it will be moderated by James Maguire, eWEEK’s Editor-in-Chief.

We’ll discuss – using X, formerly known as Twitter – challenges, issues, and best practices for governing artificial intelligence, which offers vast potential but also poses serious problems for companies that fail to properly govern this powerful emerging technology.

See below for:

  • Participant list for this month’s eWeek Tweetchat on Governing AI
  • Questions we’ll discuss in this month’s eWeek Tweetchat
  • How to Participate in the Tweetchat
  • Tentative Schedule: Upcoming eWeek Tweetchats

Participants List: Governing AI

The list of experts for this month’s Tweetchat currently includes the following – please check back for additional expert guests:

Tweetchat Questions: Governing AI

The questions we’ll tweet about will include the following – check back for more/revised questions:

  1. What are the challenges companies face with regulating generative AI? Why is it difficult?
  2. What are the problems that arise when companies don’t properly regulate their generative AI tools or other AI instances?
  3. What is one key overall rule or concept you’d recommend to help companies regulate their AI deployment?
  4. What about staff and AI regulation: what rule(s) should guide employee use of generative AI?
  5. Also regarding regulating staff: what role can AI training play? Thoughts on strategy for training staff to better regulate company AI?
  6. Which staff members should be responsible for regulating enterprise AI? A C-suite executive – which one? Does the task require a new hire?
  7. Strategies for working with vendors? How should a focus on regulating AI inform your dealings with vendors?
  8. Will the AI sector regulate itself? Or will AI regulation be government vs. industry conflict in the years ahead?
  9. Do you believe that AI will be effectively regulated at the national / international level?
  10. Overall, what’s your sense of the future of AI regulation, either at the company or higher levels?

How to Participate in the Tweetchat

The chat begins promptly at 11 AM PT on January 16. To participate:

  1. Open Twitter in your browser. You’ll use this browser to Tweet your replies to the moderator’s questions.

2. Open Twitter in a second browser. On the menu to the left, click on Explore. In the search box at the top, type in #eweekchat. This will open a column that displays all the questions and all the panelists’ replies.

Remember: you must manually include the hashtag #eweekchat for your replies to be seen by that day’s tweetchat panel of experts.

That’s it — you’re ready to go. Be ready at 11 AM PT to participate in the tweetchat.

NOTE: There is sometimes a few seconds of delay between when you tweet and when your tweet shows up in the #eWeekchat column.

#eWEEKchat Tentative Schedule for 2024*

January 16: Governing Generative AI
February 13: Data Analytics Best Practices
March 12: AI in the Enterprise: LLMs to Security
April 16: Managing Multicloud Computing
May 14: Optimizing Generative AI
June 11: Mid-Year Look Ahead: Future of Tech

*all topics subjects to change

The post eWEEK TweetChat, January 16: Governing Generative AI appeared first on eWEEK.

Top 150+ Artificial Intelligence (AI) Companies 2024 https://www.eweek.com/artificial-intelligence/ai-companies/ Mon, 25 Dec 2023 13:00:50 +0000 https://www.eweek.com/?p=222323 Artificial intelligence companies are riding a hyper-accelerated growth curve. The stunning debut of ChatGPT in November 2022 was the crack of a starting gun — the platform attracted 100 million users within months. The world woke up to the vast potential of AI, particularly generative AI. But in truth, AI companies have enjoyed huge investments […]

The post Top 150+ Artificial Intelligence (AI) Companies 2024 appeared first on eWEEK.

Artificial intelligence companies are riding a hyper-accelerated growth curve. The stunning debut of ChatGPT in November 2022 was the crack of a starting gun — the platform attracted 100 million users within months. The world woke up to the vast potential of AI, particularly generative AI.

But in truth, AI companies have enjoyed huge investments for years. Businesses have lavished money on machine learning, automation, robotics, and AI-based data analytics — even generative AI tools. The algorithm has become the foundational technology of business.

To chronicle this growth, this list of AI companies reflects the chaotic and moment-by-moment shifts disrupting the tech industry. It covers the full ecosystem of AI vendors: new generative AI companies, entrenched giants, AI purveyors across verticals, and upstart visionaries with a gleam in their eyes.

There’s no telling which of these cohorts will most influence AI’s future. Artificial intelligence is like no technology before it; it’s the first technology in history that can evolve without human assistance, and so is wildly unpredictable.

Yet while many of these AI companies won’t survive, the players on this list — as a whole — will profoundly reshape technology, not to mention education, the arts, retail, and the entirety of culture.

The end result of it all? Let’s keep our fingers crossed.

Top AI Companies

Jump to the category:

AI Giants

AI Pioneers

AI Visionaries

Generative AI Companies

AI Enterprise Majors

AI Robotics and Automation Companies

Conversational AI Companies

Healthcare AI Companies

Financial AI Companies

Education AI Companies

Cybersecurity AI Companies

Retail AI Companies

AI Industry Organizations

The Bottom Line: AI Companies

AI Giants

It’s no coincidence that this top AI companies list is comprised mostly of cloud providers. Artificial intelligence requires massive storage and compute power at the level provided by the top cloud platforms.

Additionally, these cloud leaders all offer a growing menu of AI solutions to their existing clients. This gives them an enormous competitive advantage in the battle for AI market share. Furthermore, the cloud leaders all have deep pockets, and AI development is exceptionally expensive.

Microsoft icon.


As a dominant provider of enterprise solutions and a cloud leader — its Azure Cloud is second only to AWS — Microsoft is investing heavily in AI. For example, it has significantly expanded its relationship with OpenAI, the creator of ChatGPT. Leveraging its massive supercomputing platform, its goal is to enable customers to build out AI applications on a global scale. It’s likely that Microsoft will be the leading provider of AI solutions to the enterprise.

Amazon Web Services icon.

Amazon Web Services

As the top dog in the all-important world of cloud computing, few companies are better positioned than AWS to provide AI services and machine learning to a massive customer base. In true AWS fashion, its profusion of new tools is endless and intensely focused on making AI accessible to enterprise buyers. AWS’s long list of AI services includes quality control, machine learning, chatbots, automated speech recognition, and online fraud detection.
eWeek video: AWS VP Bratin Saha on the Bedrock Generative AI Tools

Google icon.


The search giant’s historic strength is in algorithms, which is the very foundation of AI. Though Google Cloud is perennially a distant third in the cloud market, its platform is a natural conduit to offer AI services to customers. Demonstrating its competitive focus on AI, Google rolled out the AI platform Bard soon after OpenAI debuted ChatGPT. It’s a safe bet that Google will be a leader in AI in the years ahead.

IBM icon.


A top hybrid and multicloud vendor, boosted by its acquisition of Red Hat in 2019, IBM’s deep-pocketed global customer base has the resources to invest heavily in AI. IBM has an extensive AI portfolio, highlighted by the Watson platform, with strengths in conversational AI, machine learning, and automation. The company invests deeply in R&D and has a treasure trove of patents; its AI alliance with MIT will also likely fuel advances.
eWeek feature: IBM Think 2023: AI and Quantum Computing

Nvidia icon.


All roads lead to Nvidia as AI grows ever more important. At the center of Nvidia’s strength is the company’s wicked-fast GPUs, which provide the power and speed for compute-intensive AI applications. Additionally, Nvidia offers a full suite of software solutions, from generative AI to AI training to AI cybersecurity. It also has a network of partnerships with large businesses to develop AI and frequently funds AI startups.
eWeek video: Nvidia CSO David Reber on AI and Cybersecurity

Meta icon.


Meta — the parent company of Facebook, Instagram, and many other popular platforms — has had a slightly slower start on generative AI than some of the other tech giants, but it has nonetheless blazed through to create some of the most ubiquitous and innovative solutions on the market today. Meta’s Llama, for example, is one of the largest and easiest to access LLMs on the market today, as it is open source and available for research and commercial use. The company is also very transparent with its own AI research and resources.

Baidu icon.


Little known in the U.S., Baidu owns the majority of the internet search market in China. The company’s AI platform, Baidu Brain, processes text and images and builds user profiles. Baidu has announced plans to use its AI technology to create an autonomous ride-hailing service. It has also launched its own ChatGPT-like tool, a generative AI chatbot called Ernie Bot.

Oracle icon.


Oracle’s cloud platform has leapt forward over the past few years — it’s now one of the top cloud vendors — and its cloud strength will be a major conduit for AI services. To bulk up its AI credentials, Oracle has partnered with Nvidia to boost enterprise AI adoption. The company stresses its machine learning and automation offerings and also sells a menu of prebuilt models to enable faster AI deployment.
eWeek video: Oracle Cloud’s Leo Leung on Cloud Challenges and Solutions

Alibaba Cloud icon.


Alibaba, a Chinese e-commerce giant and leader in Asian cloud computing, announced in early 2023 that it will split into six divisions, each empowered to raise capital. Of particular note is the newly formed Cloud Intelligence Group, which handles cloud and AI. Notably, Alibaba’s CEO will lead this group. Alibaba has been greatly hampered by government crackdowns, but early news reports suggest this new formation is in keeping with government wishes, allowing the Cloud Intelligence Group to grow its AI rapidly. The company is also developing a ChatGPT-like tool.

Also see: Top Generative AI Apps and Tools

AI Pioneers

Think of these AI companies as the forward-looking cohort that is inventing and supporting the systems that propel AI forward. It’s a mixed bunch with diverse approaches to AI, some more directly focused on AI tools than others.

These companies are at the center of a debate in the tech industry: which group of companies will have the most control over the future of AI?

Will it be these pioneers, these agile and innovative players? Or will it be the giant cloud vendors (see above) that have the deep infrastructure that AI needs and can sell their AI tools to an already-captive customer base?

The smart money bets on the cloud players, but it remains an open question.

By the way, note when most of these pioneer companies were founded: roughly between 2009 and 2013, a fertile time to launch a data or AI initiative — and long before the ChatGPT hype cycle.

Also see: Generative AI Companies: Top 12 Leaders 

OpenAI icon.


The world was forever changed when OpenAI debuted ChatGPT in November 2022 — a major milestone in the history of artificial intelligence. Founded in 2015 with $1 billion in seed funding, San Francisco-based OpenAI benefits from a cloud partnership with Microsoft, which has invested $13 billion in OpenAI. Not content to rest on its success, OpenAI launched GPT-4, a larger multimodal version of its successful LLM foundation model. The company also offers DALL-E, which creates artistic images from user text prompts.

C3 AI icon.


Founded in 2009, C3.ai is part of a new breed of vendors that can be called an “AI vendor”: not a legacy tech company that has shifted into AI but a company created specifically to sell AI solutions to the enterprise. The company offers a long menu of turnkey AI solutions so companies can deploy AI without the complexity of building it themselves. Clients include the U.S. Air Force, which uses AI to predict system failure, and Shell, which uses C3.ai to monitor equipment across its sprawling infrastructure.
eWeek feature: C3.ai vs DataRobot: Top Cloud AI Platforms

H2O.ai icon.


Founded in 2011, H2O.ai is another company built from the ground up with the mission of providing AI software to the enterprise. H2O focuses on “democratizing AI.” This means that while AI has traditionally been available only to a few, H2O works to make AI practical for companies without major in-house AI expertise. With solutions for AI middleware, AI in-app stores, and AI applications, the company claims 20,000 customers for its H2O Cloud.
eWeek video: H2O.ai’s Prashant Natarajan on AI and Computer Vision

DataRobot icon.


Founded in 2012, DataRobot offers an AI Cloud that’s “cloud-agnostic,” so it works with all the cloud leaders (AWS, Azure, and Google). It’s built with a multicloud architecture that offers a single platform accessible to all manner of data professionals. Its value is that it provides data pros with deep AI support to analyze data, which supercharges data analysis and processing. Among its outcomes is faster and more flexible machine learning model creation.
eWeek feature: DataRobot vs. H2O.ai: Top Cloud AI Platforms

Snowflake icon.


Founded in 2012, Snowflake is a next-gen data warehouse vendor. Artificial intelligence requires oceanic amounts of data, properly prepped, shaped, and processed, and supporting this level of data crunching is one of Snowflake’s strengths. Operating across AWS, Microsoft Azure, and Google Cloud, Snowflake’s data cloud aims to eliminate data silos for optimized data gathering and processing.
eWeek video: Snowflake’s Torsten Grabs on AI and Democratizing Data

Dataiku icon.


Founded in 2013, Dataiku is a vendor with an AI and machine learning platform that aims to democratize tech by enabling both data professionals and business professionals to create data models. Using shareable dashboards and built-in algorithms, Dataiku users can spin up machine learning or deep learning models; most helpfully, it allows users to create models without writing code.

RapidMiner icon.


An enterprise-grade data science platform, RapidMiner includes a no-code AI app-building feature that allows non-technical users to create applications without writing software; it also offers a no-code MLOps solution that uses a containerized approach. As a sign of the times, users can build models using a visual, code-based, or automated approach. Founded in 2007, RapidMiner was acquired in 2022 by Altair, a publicly traded IT company that provides a wide range of enterprise tech services.

Domino Data Lab icon.

Domino Data Lab

Founded in 2013, the Domino Cloud is a fully managed MLOps (machine learning operations) offering that supports scalable enterprise data science development. Notably for its enterprise customers, the company’s open-source platform can create and train generative AI models. Domino Data Lab has partnered with Nvidia to provide a faster development environment.
eWeek video: Domino Data Lab’s Jack Parmer on “Code First” Data Science

Databricks icon.


Founded in 2013, Databricks offers an enterprise AI cloud platform that supports the flexible data processing needed to create AI and ML deployments. Think of this data solution as the crucial building block of artificial intelligence. Databricks ingests and preps data from myriad sources; its data management and data governance tools work with any of the major cloud players. The company touts its integration of the data warehouse (where the data is processed) and the data lake (where the data is stored).
eWeek video: Databricks’s Chris D’Agostino on AI and Data Management

Adobe icon.


Adobe is a SaaS company that primarily offers marketing and creative tools to its users. The company has begun to enhance all of these products with Sensei, a robust generative AI tool and assistant that personalizes marketing assets, offers smarter and more detailed customer analytics, edits visual assets for better quality, makes predictions and forecasts for optimal advertising campaigns, and creates documents through content intelligence and smart form field recognition. Beyond Sensei, Adobe also offers Adobe Firefly, a newer tool that enables users to create images and image effects with text-based inputs.

Alteryx icon.


A prime example of a mega theme driving AI, Alteryx’s goal is to make AI models easier to build. The goal is to abstract the complexity and coding involved with deploying artificial intelligence. The platform enables users to connect data sources to automated modeling tools through a drag-and-drop interface, allowing data professionals to create new models more efficiently. Users grab data from data warehouses, cloud applications, and spreadsheets, all in a visualized data environment. Alteryx was founded in 1997.
eWeek video: Alteryx’s Suresh Vittal on the Democratization of Data Analytics

Inflection AI icon.

Inflection AI

Inflection AI labels itself as an AI studio that is looking to create advanced applied AI that can be used for more challenging use cases, like more fluent human-to-computer direct communication. While it has hinted at other projects in the works, its primary product right now is Pi, a conversational AI that is designed to take a personalized approach to casual conversations. Pi can be customized and used on iMessage, WhatsApp, Instagram, and Facebook. The company is run by many former leaders from DeepMind, Google, OpenAI, Microsoft, and Meta.

Scale AI icon.

Scale AI

Scale is an AI company that covers a lot of ground with its products and solutions, giving users the tools to build, scale, and customize AI models — including generative AI models — for various use cases. The Scale Data Engine simplifies the process of collecting, preparing, and testing data before AI model development and deployment, while the Scale Generative AI Platform and Scale custom LLMs give users the ability to fine-tune generative AI to their specifications. Scale is also a leading provider of AI solutions for federal, defense, and public sector use cases in the government.

Arista icon.

Arista Networks

Arista Networks is a longstanding cloud computing and networking company that has quickly advanced its infrastructure and tooling to accommodate high volume and frequency AI traffic. More specifically, the company has worked on its GPU and storage connections and sophisticated network operating software. Tools like the Arista Networks 7800 AI Spine and the Arista Extensible Operating System (EOS) are leading the way when it comes to giving users the self-service capabilities to manage AI traffic and network performance.

Cloudera icon.


Having merged with former competitor Hortonworks, Cloudera now offers the Cloudera Data Platform and the Cloudera Machine Learning solution to help data pros collaborate in a unified platform that supports AI development. The ML solutions perform data prep and predictive reporting. As an example of emerging trends, Cloudera provides “portable cloud-native data analytics.” Cloudera was founded in 2008.
eWeek video: Cloudera’s Ram Venkatesh on the Cloudera Roadmap

Accubits icon.


Accubits is a blockchain, Web3, and Metaverse tech solutions provider that has expanded its services and projects into artificial intelligence as well. The company primarily works to support other companies in their digital transformation efforts, offering everything from technology consulting to hands-on product and AI development. The company’s main AI services include support for AI product and model development, consulting for generative AI projects, solution architecting, and automation solutions.

Also see: Generative AI Startups 

And: Best Machine Learning Platforms

AI Visionaries

If the AI pioneers are a mixed bag, this group of AI visionaries is heading off in an even wider blend of directions. These AI startups are closer to the edge, building a new vision even as they imagine it — they’re inventing the generative AI landscape in real time. More than any technology before, there’s no roadmap for the growth of AI — yet these generative AI startups are proceeding at full speed.

Adept icon.


Currently, generative AI platforms like DALL-E and GPT-4 create images or text in response to user text prompts. Adept is building the next step: It’s creating a full-fledged digital assistant — “an AI teammate for everyone” — that will execute a series of complex commands based on text prompts. For example, if you type in the prompt “convert this client into a sales opportunity,” the Adept digital assistant performs various actions to complete the sale. Ideally, Adept’s platform will be able to use any API, software app, or website just as a human would. Though Adept is a fledgling — founded in 2022 — it’s already attracted $400 million in funding.

Synthesia icon.


Is the person in this video real or virtual? Synthesia uses AI to create video avatars who speak and present as if they’re human. The AI company offers more than 150 stock AI avatars to allow users to create a virtual talking head using text prompts. To add realism, the avatars can be customized with facial gestures like raised eyebrows, head nods, and local languages and dialects.
eWeek video: Synthesia CEO Victor Riparbelli on AI and Video Avatars

Ironclad icon.


Ironclad is a contract lifecycle management vendor that uses AI to manage contract data, contract creation, analytics, and more. Its contract review process is thorough and customizable, offering users AI-driven suggestions for how to improve existing contracts based on both best practices and the AI playbooks users upload themselves; the platform also includes a comprehensive AI-powered editor and a repository that makes contracts editable in a Word-Document-like format. More recently, the vendor has come out with Ironclad Contract AI, an AI assistant that supports users with chat-driven solutions for additional contract tasks and queries.

Cohere icon.


Founded in 2019 by an elite group of AI experts, most of whom were former researchers at Google Brain, Cohere’s goal is to enable more natural communication between humans and machines. The startup builds large language models for enterprise customers, accessible via an API, which is clearly a lucrative new niche. Funding has gushed in — the company is now valued at more than $2 billion — and Google has partnered with Cohere, providing deep infrastructure support.

Abacus.ai icon.


The Abacus platform offers a generative AI service that enables clients to create synthetic data to complement their existing data sources. Synthetic data is data created by artificial intelligence instead of actual events; it’s particularly useful in building machine learning models. Founded in 2019, Abacus creates pipelines between data sources — such as Google Cloud, Azure, and AWS — and then allows users to custom-build and monitor machine learning models.

Anthropic icon.


Founded by two former senior members of OpenAI, Anthropic’s generative AI chatbot, Claude, provides detailed written answers to user questions. In essence, it’s another tool that operates like ChatGPT. But while ChatGPT’s parent company, OpenAI, is funded heavily by Microsoft, Anthropic has benefitted from a $300 million investment from Google. Anthropic claims that Claude is less prone to produce harmful material than ChatGPT.

Glean icon.


Considered one of the unicorns of the emerging generative AI scene, Glean provides AI-powered search that primarily focuses on workplace and enterprise knowledge bases. With its Workplace Search, Assistant, Knowledge Management, Work Hub, and Connectors features, business leaders can set up a self-service learning and resource management tool for employees to find important documentation and information across business applications and corporate initiatives.

Gong icon.


Gong is a fast-growing provider of customer service, sales, and marketing solutions that focus on revenue and engagement intelligence and analytics. AI is infused throughout the platform and is used to provide contextual information and recommendations for customer interactions, as well as coaching for internal team members. The vendor also offers its Smart Trackers tool, which gives users the ability to train Gong’s AI to more granularly detect certain types of customer interactions and red-flag behaviors.

Runway icon.


The three founders of Runway met in art school, where they were immersed in digital design software. Their generative AI platform, which is browser-based and requires no plugins, creates images and videos from text prompts. Think of it as a filmmaker’s dream: If you can imagine it, the Runway platform will help you create it. Runway already has a major production credit for the film Everything Everywhere All At Once, which won Best Picture in the 2023 Academy Awards.

Openstream.ai icon.


Openstream.ai is a player in the rapidly growing conversational AI market. Openstream.ai’s Eva platform leverages sophisticated knowledge graphs that use both structured and unstructured data. This mix is important because the data harvested from social media networks is unstructured. Openstream.ai uses this AI architecture to power natural language understanding (NLU), which involves levels of reading comprehension.

Samsara icon.


Samsara is an IoT company that has brought forth several innovative technologies over the years, but more recently, it has expanded into AI for driver and road safety. The company’s built-in AI and advanced edge computing for vehicles give drivers and/or fleet managers real-time insights into road conditions and driving performance, as well as coaching workflows and in-cab driver assistance. AI dash cams are built into vehicles and designed to send footage directly to the cloud, so fleet managers and business owners can review driver and vehicle issues in a timely manner.

Moveworks icon.


Moveworks is an AI company that focuses on creating generative AI and automated solutions for business operations and employee and IT support. The platform is filled with AI-powered features, including AI workflows, analytics, knowledge management, and ticket and task automation. The company is also leading the way with copilot assistive AI technology, giving users access to tools like MoveLM, an LLM that’s dedicated to employee support queries and tasks.

Synthesis AI icon.

Synthesis AI

Synthesis AI is a generative AI and synthetic data company that focuses on creating data and models for computer vision use cases. The platform can be used for a variety of use cases spanning across industries, including AR/VR/XR, virtual try-on, teleconferencing, driver and pedestrian monitoring, and security. Its primary products are Synthesis Humans and Synthesis Scenarios.

Insitro icon.


Founded by a former professor of machine learning at Stanford, Insitro’s goal is to improve the drug discovery process using AI to analyze patterns in human biology. Drug discovery is enormously expensive, with low success rates, so AI’s assistance is greatly needed. Driving this development is the company’s mixed team of experts, including data scientists, bioengineers, and drug researchers.

Eightfold AI icon.

Eightfold AI

Eightfold AI is a vendor that uses AI-powered technology to make recruitment, onboarding, retention, and other organizational talent management tasks easier to manage at scale. Users can work with the vendor’s all-encompassing Talent Intelligent Platform, which includes features not only for talent acquisition and talent management but also for resource management. Its automations and smart analytics help users to comb through larger quantities of applicants at a quicker pace while ensuring they identify top talent and new talent pipelines with minimal bias.

InVideo icon.


InVideo is an AI video company that focuses on automating script, scene, voiceover, and overall video production. The platform is frequently used for digital marketing and content marketing projects, allowing users to transform blogs and other text prompts into YouTube, talking avatar, Instagram, and other types of engaging video content. Users can customize the content the platform generates by inputting target audience, platform, and other customization instructions.

FarmWise icon.


Forget using chemicals to kill weeds in agricultural fields: FarmWise’s weeding robot uses AI and computer vision to yank out weeds without the herbicide. The FarmWise machine resembles a tractor with many arms and uses what the company calls its Intelligent Plant Scanner, a tool that is capable of sub-inch weeding accuracy.

Also see: Generative AI Examples

Generative AI Companies

Generative AI is a type of artificial intelligence that can generate content based on user text prompts. The benefits of generative AI are remarkable: finished essays, interesting graphics, complex software code, and the list goes on. At worst, generative AI can lead to cybersecurity concerns or “hallucinate,” meaning it creates false or even defamatory information. Despite these challenges, businesses are flocking to the new technology; it promises massive disruption at levels we can’t yet fully predict. Meanwhile, generative AI startups are launching daily.

Also, a highly charged debate is roiling within the generative AI sector: these AI platforms are trained on a massive store of existing material, including the work of artists and writers. But what are the copyright issues? Who “owns” the output of generative AI applications? These are thorny ethical issues with no clear answer at this point.

Also see: ChatGPT vs. GitHub Copilot

And: ChatGPT vs. Google Bard

Tabnine icon.


Tabnine is an AI company that focuses on providing AI assistance for coding and product development. The tool is designed to automate and complete code wherever possible, provide coding suggestions, and do all of this work while also ensuring that all code and data remains secure and compliant. The tool emphasizes AI ethics as well, ensuring users know that it has only been trained on open-source data repositories with permission.

Rephrase.ai icon.


This generative AI platform is a text-to-video studio. It turns your prompts into videos with digital avatars. To help marketing efforts, the solution then assists in monitoring your outreach efforts after you publish your video. Rephrase.ai uses AI to “learn” people’s facial patterns to help make their videos more realistic.

Midjourney icon.


A generative AI service that creates images from natural language text prompts, Midjourney is one of the most popular generative AI tools. Founded in 2022, it has already been used to generate surprisingly high-profile art: the English publication The Economist used it to create its cover image, and a Midjourney image scored top honors in a digital art contest hosted by the Colorado State Fair.

Infinity.ai icon.

Infinity AI

Infinity AI speeds up the process of building digital models by employing AI to create and shape synthetic data (synthetic data is computer-generated data churned out to fill in a model). In essence, Infinity AI uses AI to offer synthetic data-as-a-service, which is a niche sector that will grow exceptionally quickly in the years ahead.

Notion icon.


Notion is a project management platform that has pioneered AI assistance tools for project management professionals. Its latest collection of features, Notion AI, is available directly inside of Notion for users who want to optimize and automate their project workflows. Notion’s AI assistance can be used for task automation, note and doc summaries, action item generation, and content editing and drafting.

Podcast.ai icon.


Who needs humans? Podcast.ai is a podcast created by generative AI. Each episode is produced using realistic voice models, and the text is culled from archival material about that guest. The company released a Steve Jobs “appearance” by feeding the system his biography and reams of related material; the real-life Joe Rogan interviewed “Steve Jobs.”

Hugging Face icon.

Hugging Face

Originally the developer of a chatbot aimed at the teen crowd, Hugging Face has evolved into a repository for prebuilt machine learning models. Now a significant player in the generative AI AI sector, thousands of companies use Hugging Face’s platform to generate AI-based applications. The company’s motto is “The AI community building the future.”

Stability.ai icon.

Stability AI

Stability AI is a brand new generative AI company that supports Stable Diffusion, an AI model that generates images in response to user text prompts. Notably, Stability AI offers StableLM, an entire group of language models. Given that large language models are the very foundation of generative AI, Stability AI is certainly playing a role in developing this new technology.

Mostly.ai icon.


Focusing on the synthetic data sector, MOSTLY AI touts that the synthetic data it creates with generative AI appears as authentic as actual consumer data. The advantage is that this data doesn’t contain the original private data, so it’s compliant with privacy and data governance standards. The company works across a range of industries, including banking and insurance.

Syntho icon.


Syntho’s Syntho Engine 2.0 uses generative AI to create synthetic data, offering a self-service platform. The company creates data to build digital twins that respect privacy and GDPR regulations. Its goal is to “enable the open data economy,” in which data can be shared more widely while ensuring sensitive consumer data is protected.

Jasper icon.


Similar to ChatGPT, though with a marketing focus, Jasper uses generative AI to churn out text and images to assist companies with brand-building content creation. The AI solution learns to create in the company’s “voice,” no matter how mild or spiky, for brand consistency. The company also claims to incorporate recent news and information for a current focus on any market sector.

Biomatter icon.


Biomatter leverages generative AI to create synthetic biologic materials, specifically new proteins “for health and sustainable manufacturing applications.” This technology for creating synthetic proteins means new enzymes can be created with completely novel properties and use cases. Clearly, this is just one of many examples of how generative AI will play a crucial role in the future of medicine.

You.com icon.


Should Google be threatened in its internet search business? If so, the generative AI platform You.com — “the AI search engine you control” — could be part of the competition. Type a query into You.com, and the ChatGPT-style website will create content based on your request. By the way, you’ll note that You.com’s homepage looks remarkably like Google’s.

Osmo icon.


Computers, it seems, will soon have a sense of smell. Osmo is digitizing and analyzing scents with the goal of improving healthcare and consumer products like shampoo and insect repellent. The company is creating a vast “map” of scents, called a Principal Odor Map. There are said to be billions of molecules that carry a scent, but only about 100 million of them are known. Osmo utilizes Google Cloud’s AI platform for its generative AI work.

Also see: Top Generative AI Apps and Tools

And: The Benefits of Generative AI

AI Enterprise Majors

A popular saying has emerged among IT experts: “Every company is a tech company.” Using technology is now so central to being competitive that it’s a core focus for every company, regardless of sector.

Now this saying has a companion: “Every tech company is an AI company.” This means that major enterprise tech vendors that have long sold legacy hardware and software are now shifting to artificial intelligence. These big vendors are using their deep pockets and expertise to create AI solutions or acquire AI companies.

In fact, these enterprise majors started investing in AI long before ChatGPT burst onto the scene. So while their tools don’t get the buzz of DALL-E, they do enable staid legacy infrastructures to evolve into responsive, automated, AI-driven platforms.

Salesforce icon.


Not long after OpenAI debuted ChatGPT, Salesforce followed up with Einstein GPT, which it calls “the world’s first generative AI platform for CRM.” Powered by OpenAI, the solution creates personalized content across every Salesforce cloud. For instance, it uses generative AI with Slack to offer conversation summaries and writing help. Also, Salesforce Ventures announced a new $250 million Generative AI Fund to invest in promising startups.
eWeek video: Salesforce Chief Scientist Silvio Savarese on Conversational AI

BMC icon.

BMC Software

Among its other AI-enhanced offerings, BMC’s Helix solution uses AI and ML-based intelligent automation as part of an IT services and operation management platform. The company also provides AIOps solutions (AI for IT operations), a sector that is evolving toward AI for overall business support. The company’s larger focus — one that relies heavily on AI — is the autonomous digital enterprise.
eWeek video: BMC CEO Ayman Sayed on DataOps and the Autonomous Digital Enterprise

Hewlett Packard Enterprise icon


HPE’s Greenlake is an IT-as-a-service solution with a hybrid cloud focus. Part of this on-demand platform is a GPU offering that enables the rapid deployment of AI and machine learning tools. HPE focuses on providing AI geared for various verticals, from healthcare to financial services to manufacturing.
eWeek video: HPE Greenlake SVP Keith White on Change in the IT Sector

Dell icon.


Dell’s APEX solution, which includes multicloud management and a SaaS-based IT services panel, enables companies to build AI-based tools ranging from fraud detection to natural language processing to recommendation engines. The company also stresses the AI support provided by its hardware, like its PowerEdge servers and PowerScale Storage.
eWeek video: Dell APEX’s Chad Dunn on Handling Multicloud Challenges

SAP icon.


The ultimate legacy software player, known for its strength in ERP, SAP has clearly moved into the AI era. Its menu of enterprise AI solutions ranges from an AI chatbot to a platform that helps companies incorporate AI into enterprise applications. For its offering of pre-trained AI models, SAP stresses compliance and transparency, which is particularly important for large enterprise clients.
eWeek video: SAP’s Irfan Khan on ‘Analytics Everywhere’

ServiceNow icon.


An enterprise leader in IT service management (ITSM), the ServiceNow AI offerings include a predictive analytics platform that supports AI tool delivery without data science experience. This is an example of the “democratization of tech,” in which the levers of tool creation are now open to non-tech staff. ServiceNow also provides natural language processing tools, ML models, and AI-powered search and automation.
eWeek video: ServiceNow’s Matt Schvimmer on Accelerating Cloud Migration

Broadcom icon.


Broadcom has a unique profile in the enterprise IT industry: The company supplies both semiconductors and enterprise infrastructure software; it serves markets from the data center to wireless; it even makes a play in the multicloud sector. In keeping with this broad approach, Broadcom fuels the AI market on multiple levels, notably in its generative AI business, which the company announced in March 2023 is poised to quadruple.
eWeek video: Broadcom’s Ganesh Janakiraman on Multicloud Challenges

SAS icon.


A leader in data analytics and business intelligence, SAS’s AI menu extends from machine learning to computer vision to NLP to forecasting. Notable tools include data mining and predictive analytics with embedded AI, which boosts analytics flexibility and scope and allows an analytics program to “learn” and become more responsive over time.
eWeek video: SAS’s Katy Salamati on Data and Intelligent Decisioning

Rockwell Automation icon.

Rockwell Automation

Rockwell serves the rapidly expanding market for large-scale industrial automation, including factories and other major production facilities. It has a particular strength in providing automation for edge computing deployments. In keeping with a powerful trend sweeping the AI and automation sector, Rockwell’s FactoryTalk Analytics LogixAI solution enables non-technical staff to access machine learning tools.

Informatica icon.


Founded in 1993 to serve the nascent ETL (extract, transform, and load) big data market for enterprise customers, Informatica’s current strategy involves using AI to improve data analytics and data mining for competitive value. The company’s CLAIRE solution uses repositories of metadata to fuel its AI and ML development.

Infosys icon.


Infosys touts its AI and Automation Services teams as a solution to provide AI and automation consulting, create bespoke AI platforms, and offer prebuilt cognitive modeling solutions. These include robotic process automation (RPA) tools and AI chatbot models.  The company is considered a leader in intelligent automation.
eWeek video: Infosys Consulting CEO Andrew Duncan on Tech Headwinds

Also see: Cloud and AI Combined: Revolutionizing Tech

On a related topic: The AI Market: An Overview

AI Robotic Process Automation Companies

The fields of robotics and automation existed long before AI became a viable business solution. However, early uses of robotics (notably in auto factories) were merely devices programmed to perform the same task again and again.

The more recently developed field of robotic process automation (RPA) makes full use of AI. RPA vendors develop AI-based software that learns and automatically performs routine office productivity tasks. For instance, an office manager who has to gather files for a weekly report can set up an RPA automation to do that routine task so they can focus on higher-value work.

While many large companies offer RPA as part of their overall portfolio — notably SAP, ServiceNow, and IBM — the following vendors specialize in creating intelligent automation to boost productivity.

Also see: Top Robotics Startups

UIPath icon.


Generally acknowledged as the leader in the RPA market, UiPath offers a broad suite of business automation tools across API integration, intelligent text processing, and low code app development. The company’s Marketplace platform offers an extensive menu of prebuilt automations, from “extract data from a document” to “OpenAI” to “Microsoft Office 365.”

Nuro icon.


Nuro is a robotics-focused company that uses AI, advanced algorithms, and other modern technology to power autonomous, driverless vehicles for both recreational and business use cases. The Nuro Driver technology is trained with advanced machine learning models and is frequently quality-tested and improved with rules-based checks and a backup parallel autonomy stack. The company partners with some major retailers and transport companies, including Walmart, FedEx, Kroger, and Uber Eats.

Automation Anywhere icon.

Automation Anywhere

A player in the all-important cloud native ecosystem, Automation Anywhere’s AARI tool democratizes automation by enabling non-technical staffers to create workflow automations. In 2021, the company acquired process intelligence vendor FortressIQ to expand its tool sets, which should benefit Automation Anywhere as the RPA market evolves toward more sophisticated automation.
eWeek video: Automation Anywhere CEO Mihir Shukla on Intelligent Automation

Anduril icon.


Anduril is a leading U.S. defense technology company that creates autonomous AI solutions and other autonomous systems that are primarily powered by Lattice. The tools offered by Anduril can be used to monitor and mitigate drone and aircraft threats as well as threats at sea and on land. Its most impressive autonomous systems include underwater vehicles and air vehicles for managed threat defense.

BluePrism icon.

SS&C Blue Prism

Acquired by financial services software vendor SS&C in 2022, Blue Prism appears to have enlarged its strategy from RPA to overall business automation. This is very much in keeping with the industry shift toward more all-encompassing automation: As AI gets smarter, RPA systems accomplish that much more. Included in the Blue Prism offering are tools that perform ML decisioning and process orchestration.

EdgeVerve icon.


EdgeVerve serves its enterprise clients a growing menu of pre-fabricated automations to speed up workflows in the most important and commonly needed business areas. Products include Finacle Treasury for banking and TradeEdge for supply chain management. Like the rest of the RPA sector, EdgeVerve is evolving its automation capabilities to support digital transformation; in essence, we’re heading toward a world where the office runs itself. Infosys acquired EdgeVerve in 2014.

Shield AI icon.

Shield AI

Shield AI is an innovative AI startup that has quickly gained notoriety and capital for its AI pilot technology. Hivemind is an AI pilot that can fly aircraft in both commercial and battle settings, giving users greater insights into their locations and travel paths as well as what’s happening with other pilots and aircraft in their fleet. At this point, Shield AI’s technology is powering several of the vendor’s own intelligent aircraft, including jets, V-BAT teams, and Nova 2.

WorkFusion icon.


RPA software platforms create “digital workers,” otherwise known as AI-powered software robots. WorkFusion builds on this with a platform that includes six digital staffer personas. Each category of virtual worker is geared for the most common and/or important automation scenario. WorkFusion has a strong presence in the financial sector.

NICE icon.


A strong contender in the call center market, NICE’s RPA solutions are geared toward an array of customer-facing support functions. Significantly, its tool set includes speech and sentiment analysis, which is critical to the retail environment because it can effectively understand the emotions of callers. This helps an agent respond accordingly — this type of sentiment analysis is a particularly hot area in the AI market. Also helpful, the company’s NEVA Discover tool aims to calculate the ROI of potential automations.

Pega icon.


As businesses seek to grow toward a more fully automated environment, Pegas’ RPA architecture has kept pace, adopting a strategy that uses real-time data to guide automated customer interactions. The company touts its ability to read customer intentions, from potential purchases to imminent cancellations, before a customer acts. Overall, the company’s strategy is geared toward greater scalability to support increasingly all-encompassing automation.

Also see: Robotic Process Automation Vendors

And: Best Machine Learning Platforms

Conversational AI Companies

We don’t want to just click on our software; we want to talk with it, and we want much easier and more natural ways to control software. Software equipped with conversational AI capabilities allows just this, as it understands and mimics human speech.

Conversational AI is powered by natural language processing, a subsector of AI focused on translating the idiosyncrasies of human speech into computer commands. There are numerous advantages to this, but here’s a big one: Conversational AI enables non-technical staff to use AI. No need for programmers or experts, everyone is invited.

On a related subject: Algorithms and AI

Gridspace icon.


Gridspace is a conversational AI solution for different businesses, giving users access to virtual AI agents, advanced analytics, and AI coaching for better conversational outcomes with customer service reps. Virtual agents can be customized to quality assurance, revenue management, lead generation, and self-service customer relationship management requirements. The company consists of a multidisciplinary team of engineers, designers, and experts from SRI Speech Labs, where Siri was developed.

Kore.ai icon.


Considered a top player in conversational AI, Kore.ai’s no-code tool set allows non-technical staff to create versatile and robust virtual assistants. This “build it yourself” ethos is a dominant theme in the AI chatbot sector. The company is also known for its extensive NLP solutions.

Cognigy icon.


A core offering of conversational AI vendors is tools that improve the performance of call center agents (or other voice-based customer reps). To serve this market, Cognigy offers Cognigy Agent Assist. The company also offers analytics tools and a low-code platform to enable users to create new bot assistants as needed for their situation.

Amelia icon.


Amelia’s intelligent agents leverage advanced NLU capabilities — essentially the leading edge of AI chatbot technology. NLU technology enables a virtual agent to use sentiment analysis, which helps reps monitor the emotions of callers. This is a leading frontier of the conversational AI market.

OneReach.ai icon.


OneReach.ai is an example of a leading trend in the conversational AI market, as the company evolves its offerings from a narrow call center focus toward an enterprise-wide “AI-based virtual staff member.” The result of this trend is that the conversational AI sector is merging with the RPA sector (see above) as conversational AI companies produce full-fledged digital team members.

Avaamo icon.


With a background in healthcare-focused conversational AI, Avaamo is extending its reach across various industry sectors. Looking ahead, this SaaS vendor has set up a waitlist for early access to its AvaamoGPT generative AI tool, which it touts as “a next-generation assistant for your enterprise.”

Yellow.ai icon.


With an intuitive user interface, Yellow.ai’s product offering includes user-friendly prefabricated models to deploy conversational AI agents; ease of use is a top priority in the conversational AI market. To help integrate third-party functionality, Yellow.ai has built a marketplace where customers can select third-party tools for specific tasks.

Boost.ai icon.


Boost.ai offers a full menu of advanced chatbot orchestration tools to speed deployment. To help call center reps boost performance with customer calls, Boost.ai provides agents with a large repository of support data. The company claims its Hybrid NLU technology improves the quality of its virtual agents.

Also see: Generative AI Examples

And: Top Natural Language Processing Companies

Healthcare AI Companies

AI healthcare companies are incentivized by two crucial advantages provided by AI and generative AI: First, artificial intelligence greatly expands the capabilities of medical professionals — and better tools are literally a matter of life and death. Second, AI is adept at streamlining bureaucracy, a huge part of the healthcare sector, thus saving time and money. Look for healthcare to be a non-flashy but very powerful driver of AI’s progress.

PathAI icon.


PathAI is one of the most advanced pathology-focused AI companies today, giving patients, laboratories, and pharmaceutical companies alike access to the AI-powered insights and solutions they need. The company offers accessible AI algorithms for optimized clinical trials, particularly for oncology, as well as AI-powered companion diagnostics, pre-screening predictions, spatial analyses, and translational research. The company’s algorithms and products specifically support biomarker quantification for various cancers, disease severity assessments, quality control, tumor cellularity quantification, and molecular prediction.

Viz.ai icon.


Viz.ai offers AI-powered platforms and applications for care coordination, ensuring patient care is handled more holistically by all of their healthcare providers. The Viz.ai One platform is specifically designed to work in different areas of healthcare, including neurology, cardiovascular, vascular, trauma, and radiology. With this platform, healthcare providers quickly receive insights, clear images, alerts, and communications from other relevant providers, making it so they can more quickly and accurately diagnose their patients. Viz.ai is available in both the U.S. and the EU.

CloudMedX icon.


Supported by AI, CloudMedx harvests data and creates portraits of patients with the goal of improving its core predictive analytics to create better healthcare results. Among the risk criteria it looks for, the company’s AI-based data processing aims to assess the extent of patients’ medical issue risks based on a given procedure.

Paige icon.

Paige AI

Paige AI is a generative AI company in the healthcare sector that focuses on pathology, specifically cancer diagnostics. Its detailed imaging technology, AI-driven workflows and recommendations, and other smart features assist healthcare professionals in breast and prostate cancer diagnosis as well as in optimizing hospital and lab operations.

Insilico Medicine icon.

Insilico Medicine

Insilico Medicine is a research and development company that uses artificial intelligence for smarter biology and chemistry research and pharmaceutical analytics. Its PHARMA.AI suite includes PandaOmics, a tool for multi-omics novel target discovery and deep biology analysis; Chemistry42, an ML-powered tool for drug design and automated novel molecule creation; and InClinico, a tool that can both design and predict the success rate of a clinical trial.

Medtronic icon.


A fascinating fact about Nvidia: if you dig deep into the AI landscape, you’ll see Nvidia again and again. A good example of this is Medtronic, which is a well-known medical device maker that operates the Genius AI solution, which enhances the detection of polyps in colonoscopies. The company has partnered with Nvidia to use AI to create a range of next-gen tools for diagnosis and treatment.

Enlitic icon.


Enlitic’s Curie platform uses artificial intelligence to improve data management in the service of better healthcare. The goal is to make data more accurate, useful, and uniform to enable doctors and other healthcare professionals to make better patient care decisions.

Deepcell icon.


Deepcell is a biotech startup — spun out of Stanford University in 2017 — that leverages AI to examine and classify cells. By identifying viable cells based on morphology (the study of shapes and arrangement of parts), Deepcell technology can more accurately perform diagnostic testing.

Activ Surgical icon.

Activ Surgical

Activ Surgical is an AI healthcare company that uses AI to provide real-time surgical insights and recommendations during surgical operations. The ActivSight product, powered by the ActivEdge platform, is designed to not only give surgeons easy-to-view real-time data but also to make it possible for them to switch between dye-free and dyed visualizations, depending on their needs. Important for healthcare workers, the solution is MIS-system compatible.

Arterys icon.


To enhance medical imaging, Arterys accesses cloud-based GPU processors, which it uses to support a deep learning application that examines and assesses heart ventricles. This AI-based automated measurement of ventricles allows healthcare professionals to make far more informed decisions.

Etcembly icon.


Etcembly is an immunotherapy company that focuses on designing new TCR therapies. Its therapies are optimized through a deep ML library of immunology expertise and computer-assisted immunotherapy engineering. The platform is designed to learn directly from the interactions of T-cells so appropriate TCR treatments can be identified and developed.

Corti icon.


There are numerous companies using AI to provide call center support, but Corti’s niche is the healthcare sector. To provide a virtual voice assistant geared for the healthcare sector, the company’s solution has been trained with countless hours of conversations between healthcare workers. Among other tasks, the solution can support QA on calls to telemedicine centers.

Butterfly Network icon.

Butterfly Network

A medical imaging vendor, Butterfly Network uses AI in myriad ways. In 2022, Butterfly Network debuted FDA-cleared AI software to support the use of ultrasound technology. In 2023, the company received FDA approval for its AI-enabled lung tool, which uses deep learning technology to more quickly and fully assess lung health.

Owkin icon.


Owkin uses AI to drive predictive analytics for the development of better drug solutions for a variety of diseases. Perhaps most notably, the company’s platform facilitates collaboration between data scientists and academic researchers. To support this development, Owkin has received a major investment from Sanofi, a French multinational pharmaceutical company.

GE Healthcare icon.

GE HealthCare

Spun off from conglomerate GE in January 2023, GE HealthCare is building a platform called the Edison AI Orchestrator. Edison is designed to fully integrate AI-enabled clinical applications into radiology for both GE and non-GE devices; this is being done to boost the quality of medical decision-making. Additionally, the company has hired a former Amazon machine learning executive to assist in its AI healthcare expansion.

Caption Health icon.

Caption Health

A maker of AI-based ultrasound guidance software, Caption Health’s software makes ultrasound exams more efficient. This small company is in the process of getting a lot more resources for growth: In February 2023, newly formed GE HealthCare (see above) announced that it is acquiring Caption Health.

Stryker icon.


Already a large and well-established medical device maker, in 2021, Stryker acquired the AI company Gauss Surgical and is aggressively moving to deploy AI more broadly across its product offerings. Among its notable products is the AI-based Stryker Mako robot, which can assist with numerous medical procedures.

Cleerly icon.


In service to Cleerly’s ambitious goal — “creating a world without heart attacks” — the company’s artificial intelligence platform performs an analysis of non-invasive coronary computed tomography angiography (CCTA) scans to assess plaque levels in the heart. Cleerly’s algorithms mine an extensive database full of lab images to compare a patient with historical records.

ClosedLoop.ai icon.


ClosedLoop’s data science platform leverages AI to manage and monitor the healthcare landscape, working to improve clinical documentation to lower out-of-network use and predict admission and readmission patterns. Impressively, the company won the CMS Artificial Intelligence Health Outcomes Challenge in 2021.

Oncora Medical icon.

Oncora Medical

Oncora Medical’s machine learning software supports healthcare professionals with numerous administrative tasks in the manner of a digital assistant. It streamlines doctors’ time by assisting in documentation, stores all notes and reports, requests additional relevant notes from healthcare providers, and creates the needed forms for clinical and invoicing uses.

Atomwise icon.


The process of drug development has historically been slow and cumbersome, often requiring years to match compounds to develop new drugs. Atomwise aims to speed this up exponentially by using a deep learning-based discovery engine to sift through its vast database (the company claims 3 trillion compounds) to find productive matches.

Also see: Generative AI in Healthcare

And: Top AI Startups

Financial Services AI Companies

It’s clear that financial services firms are actively embracing artificial intelligence. Bank of America, in a breathless note to the investment community, opined that “AI is the new electricity.”

Wells Fargo is developing a new AI chatbot called Fargo (powered by Google AI). JP Morgan has its own Artificial Intelligence Research division. Visa — like all major finance companies — uses AI extensively to fight fraud. And more fintech companies than anyone can count are hopping on the AI bandwagon.

For more information: Cloud and AI Combined: Revolutionizing Tech

Intuit icon.


Intuit is an enterprise that has focused on providing both guided and self-service finance and tax tools to users of products like TurboTax, Credit Karma, Mint, QuickBooks, and Mailchimp. The company recently released Intuit Assist, a generative AI financial assistant that is able to provide SMB leaders with smart recommendations for their financial and customer service decisions; Intuit Assist is available for TurboTax, Credit Karma, QuickBooks, and Mailchimp. Intuit also boasts an AI research program that focuses on developing and refining new AI innovations with explainable AI, generative AI, and more.

Stripe icon.


Stripe is a SaaS-based financial services company that is frequently selected for its user-friendly payment processing features. Most recently, Stripe has jumped aboard the AI train, partnering most heavily with OpenAI at this time. The two companies have something of an exchange going on right now: Stripe is using GPT-4 for smarter documentation and content summary generation in Stripe Docs, while Stripe is helping OpenAI improve its checkout process and other customer experiences through Stripe Billing, Checkout, Tax, and other Stripe products.

Capital One icon.

Capital One

Capital One is a prime example of how financial institutions are finding multiple ways to leverage artificial intelligence. The financial company’s many AI initiatives include explainable AI, which makes the loan approval process transparent; anomaly detection, which helps fight fraud; and NLP, which improves virtual assistants for customer service.

Brighterion icon.


A division of Mastercard, Brighterion serves Mastercard’s AI needs and also provides AI services to other companies. Brighterion’s AI Express offers customized AI solutions geared to the needs of financial services companies. Brighterion touts its “custom AI that’s production ready in 6-8 weeks.”

Numerai icon.


Promoting itself as “the hardest data science tournament in the world,” Numerai’s AI-enabled, open-source platform offers a way for data scientists to predict trends in the stock market and make a profit if they’re right. The business model involves using machine learning models to forecast financial megatrends. The company is supported by Union Square Ventures, which co-founded Coinbase.

Skyline AI icon.

Skyline AI

An example of how AI can be leveraged to support virtually any financial transaction, Skyline AI uses its proprietary AI solution to more efficiently evaluate commercial real estate and profit from this faster insight. Competitors in the AI-driven real estate sector include GeoPhy and Cherre, which won the Business Intelligence Group AI Excellence Award.

Ocrolus icon.


The need for AI-based automation is enormous in the financial sector because financial services firms always have oceans of metrics and data points to digest. Ocrolus enables banks and other lenders to fight fraud by automating financial document analysis. Significantly, Ocrolus’s human-in-the-loop solution maintains human experience as a core factor in document authentication.

AlphaSense icon.


Google parent Alphabet invested a stunning $100 million in AlphaSense, valuing the company at $1.8 billion. AlphaSense competes in the lucrative business data market against big players like Bloomberg. Among AlphaSense’s AI-fueled initiatives, the company is developing a solution that can summarize financial reports to more quickly reveal salient data trends.

Zest AI icon.

Zest AI

Zest AI uses AI to sift through troves of data related to borrowers with limited credit history, helping lenders make decisions with this limited data. In particular, it helps with the auto lending market, where the company claims it cuts underwriter losses by approximately 25% by better quantifying creditworthiness.

Signifyd icon.


Signifyd is a company that uses AI to create a “score” — from 0 to 1,000 — to fight fraud in the financial sector. While the trend of deploying AI to combat financial malfeasance is sweeping the industry, Signifyd claims to distinguish itself by boosting transaction approvals and dramatically lessening false declines.

HighRadius icon.


A leading player in the accounts receivable automation software sector, HighRadius uses machine learning to help with labor-intensive tasks like matching payments with invoicing and assigning credit limits. The company partners with Citibank, Bank of America, and SAP.

On a related topic: The AI Market: An Overview

And: Top AI Software

Education AI Companies

One of the great promises of AI in education is that it will provide one-on-one tutoring and coaching opportunities, which will markedly boost student performance. If this were to fully mature, AI “teachers” would provide lessons at a far lower cost than human tutors. AI can also support teachers, helping them quickly craft lesson plans and other educational resources. In any case, learning how to use AI will become a core skill for students as it becomes woven into every element of work and culture.

For more information: Best Machine Learning Platforms

Carnegie Learning icon.

Carnegie Learning

Focusing on the K-12 market, Carnegie Learning’s MATHia with LiveLab is well recognized as an advanced AI learning app. The app uses an AI-powered cognitive learning system to support math education, offering students one-on-one interactions that allow them to work at a pace that best suits them.

Century Tech icon.


CENTURY is a UK-based educational platform company that uses neuroscience to enable enhanced learning in various high school and college core topics. CENTURY uses algorithms like those at Netflix and Amazon to match previous student experiences with what they should focus on next for optimal educational progress. Additionally, the platform offloads some repetitive teaching tasks so teachers can spend more time focusing on students’ needs.

ELSA icon.


ELSA is a company that uses AI to smooth out the user experience side of learning English as a non-native speaker. Its Speech Analyze tool uses AI to analyze user speech patterns, accents, and other details in order to give feedback on possible improvements. Users can also take assessments that ELSA’s AI uses to customize courses and learning timelines that fit that particular user. ELSA is used in both corporate and educational settings.

Kidaptive icon.


Kidaptive’s “adaptive” AI technology is referenced in its name. Founded by two Stanford alumni, Kidaptive’s Adaptive Learning Platform is heavy on next-gen technology: it uses a multi-tenant cloud deployment and is supported by Hadoop. Solutions include Learner Mosaic and Leo’s Pad to support what it calls “playful, whole child development.” Kidaptive was acquired by McGraw Hill in 2021.

Amira Learning icon.

Amira Learning

Winner of Time Magazine’s Best Inventions award in 2021, Amira Learning uses an AI-powered gamified learning environment to improve reading skills. Children read aloud as Amira provides real-time support; the solution has multiple tutoring techniques to coach young readers, including offering encouragement.

Duolingo icon.


Well known for teaching foreign language acquisition (they claim more than 50 million monthly users), Duolingo uses OpenAI’s GPT-4 to create free-flowing conversations with language learners, recreating the experience of chatting with a native speaker. Here’s an impressive credential for the company: The OpenAI website hosts a page detailing a Duolingo case study.

Cognii icon.


Cognii’s VLA (virtual learning assistant) platform speaks with students in real time, providing one-on-one coaching. The goal is to transcend the limits of a multiple-choice question format and offer a wide-ranging conversation. The company’s NLP tools respond to students’ own language styles.

Querim icon.


Focusing on short-form lessons in the STEM sector, Querium’s StepWise AI tutor provides students with constant feedback as they work through challenging projects. It is designed to detect issues and provide personalized assistance. The company promotes its “AI based on the wisdom of master teachers.”

Squirrel AI icon.

Squirrel Ai Learning

Based in China, Squirrel Ai Learning uses artificial intelligence to drive adaptive learning for students at a low cost. Its focus is personalized tutoring for the K-12 sector. The company’s engineers work to break down subjects into smaller sections, enabling the AI platform to understand exactly where each student needs help.

Also see: AI Courses: Learn AI with These 10 Courses

Cybersecurity AI Companies

The challenge with creating a list of today’s AI cybersecurity companies is that every major cybersecurity company now claims to use AI. So a list of “top AI cybersecurity companies” is essentially the same as “top cybersecurity companies.”

However, the problem is that I’ve heard big doubts from industry experts about the efficacy of AI cybersecurity; these critics say that the vendors make big noises about AI, but in fact, the technology is immature.

That issue is open to debate, but one thing is certainly true: For customers of these security companies, it’s very hard — impossible? — to look under the hood and fully understand the depth and quality of a vendor’s AI.

Will a given vendor’s AI really be able to drive predictive analytics enough to block a virus before it permeates the infrastructure? Maybe or maybe not, but those doubts aren’t stopping vendors from boasting about their AI cybersecurity solutions.

Also see: Generative AI and Cybersecurity: Advantages and Challenges

CrowdStrike icon.


CrowdStrike offers XDR (extended detection and response), a growing theme in cybersecurity that makes heavy use of artificial intelligence and automation to patrol infrastructure and quickly alert admins to threats. CrowdStrike promotes its managed XDR system’s ability to use AI to close the skills gap in cybersecurity by performing the work of missing security pros.
eWeek video: CrowdStrike’s Amol Kulkarni on Trends in Cybersecurity

Zscaler icon.


Zscaler uses a powerful emerging technology in cybersecurity called zero-trust architecture, in which the permission to move through a company’s system is severely limited and compartmentalized, greatly reducing a hacker’s access. The company’s AI models are trained on a massive trove of data to enable it to constantly monitor and protect this zero-trust architecture.

SentinelOne icon.


SentinelOne’s Singularity platform is an AI-powered, comprehensive cybersecurity solution that includes extended detection and response, an AI data lake, AI threat detection, and other features for endpoint, cloud, and identity-based security needs. Most recently, SentinelOne expanded its generative AI capabilities, using generative AI for reinforcement learning and more efficient threat detection and remediation.

Abnormal Security icon.

Abnormal Security

Protecting email is a bit of a mind game: Hackers can send deceptive phishing appeals directly to every staffer in the company, so it’s likely that someone’s going to fall for the scheme. To combat this, Abnormal Security uses AI to learn the typical behavior of every employee to help block malicious entry to the perimeter. Impressively, security leader CrowdStrike has invested in and partnered with Abnormal.

Airgap icon.

Airgap Networks

Airgap Networks is an AI-driven cybersecurity company that focuses on network and threat intelligence, agentless discovery, network segmentation and microsegmentation, and zero-trust infrastructure best practices. Most recently, Airgap released ThreatGPT, a GPT-3-powered AI solution that can more effectively microsegment networks, provide deeper AI-driven security analytics, offer context on vulnerabilities detected, and more, all in plain English so less-experienced cybersecurity professionals can take action.

Vectra icon.

Vectra AI

Vectra AI’s Cognito platform uses artificial intelligence to power a multi-pronged security offensive. This includes Cognito Stream, which sends enhanced metadata to data repositories and the SIEM perimeter protection; and Cognito Protect, which acts to quickly reveal cyberattacks.

Darktrace icon.


Darktrace’s Cyber AI Loop uses a continuous loop architecture to create a constant flow of prevention, detection, response, and healing; the idea is that the AI foundation will learn with each iteration, providing more powerful cyber protection over time. The company stresses the self-learning abilities of AI, to “learn every micro interaction” in an enterprise environment.

Sophos icon.


Clearly a leader in AI-based cybersecurity long before the current AI hype cycle, the UK-based company launched Sophos Artificial Intelligence way back in 2017. This initiative focuses on developing forward-looking advances in machine learning and data for human-AI interaction and other security uses. Sophos’s deep tool set ranges from endpoint detection to encryption to unified threat management.
eWeek video: Sophos CTO Joe Levy on AI in Cybersecurity

Fortinet icon.


At the center of today’s enterprise cyber protection is the security operations center (SOC). Fortinet’s automated SOC uses AI to ferret out malicious activity that is designed to sneak around a legacy enterprise perimeter. The strategy is to closely interoperate with security tools throughout the system, from cloud to endpoints.

Palo Alto Networks icon.

Palo Alto Networks

With a strong reputation as a cybersecurity company with an advanced strategy, Palo Alto Networks’s AI-powered Prisma SASE (secure access service edge) solution is integrated with its Autonomous Digital Experience Management (ADEM) tool. The net result is that AI helps human security admins with observability across their infrastructure, which is crucial for enterprise security.

Check Point icon.

Check Point

Check Point’s Quantum Titan offers three software blades (security building blocks) that deploy deep learning and AI to support threat detection against phishing and DNS exploits. The company also focuses on IoT, with tools that apply zero-trust profiles to guard IoT devices in far-flung networks.

SecurityScorecard icon.


SecurityScorecard is a threat and risk intelligence company that provides smart security ratings, automatic vendor detection and cyber risk quantification, and other products and services to identify risks and vulnerabilities before they spiral out of control. The company recently added generative AI to its toolkit through a security ratings platform that has OpenAI’s GPT-4 as one of its foundational models. With this new feature, users don’t have to have cybersecurity or risk management experience to ask questions of the platform and receive risk management recommendations.

Blackberry icon.

Cylance AI

A division of BlackBerry, Cylance AI touts its “seventh generation cybersecurity AI.” Due to its extended lifecycle in use by clients, the AI platform has been trained on billions of cyberthreat datasets. Given its mobile credentials, Cylance is a key player in cybersecurity for the mobile IoT world, a quickly growing sector.

BigPanda icon.


Considered a leader in the AIOps sector, BigPanda uses AI to discover correlations between data changes and topology (the relationship between parts of a system). This technology works to support observability, a growing trend in infrastructure security. In essence, BigPanda uses machine learning and automation to extend the capabilities of human staff, particularly to prevent service outages.

DataVisor icon.


DataVisor deploys AI to combat fraud across many transaction types, from digital payments to fintech platforms. For instance, it monitors transactions in real time to block credit card fraud and protects ACH and Zelle payments to fight unauthorized payments. The company was dubbed a “Cool Vendor” by Gartner in 2020.

For more information: AI vs. ML

Retail AI Companies

AI in retail typically focuses on personalizing the customer experience and supporting automation and data analytics to improve the supply chain. To fully portray AI’s role in retail, this section lists both AI vendors and large retailers that deploy AI. Both groups play a crucial role in creating and enhancing the many uses for AI in retail.

Shelf Engine icon.

Shelf Engine

Shelf Engine is an AI startup with a goal to solve one of the most problematic questions in retail: What is the optimal amount of inventory to order? This question is particularly crucial for sellers of perishable goods like fruits and vegetables. Shelf Engine works to automate the stocking process so retailers can hold the optimal inventory level and so customers find what they need but stores handle only minimal waste.

Deep North icon.

Deep North

Combining computer vision with artificial intelligence, Deep North is a startup that enables retailers to understand and predict customer behavior patterns in the physical storefront. The company provides tools to use this information to improve customer experience and boost sales. Deep North is an example of how AI is evolving toward analyzing nearly every aspect of human action.

Dynanic Yield icon.


In 2019, the fast food giant acquired Dynamic Yield, an AI-powered personalization platform that has worked with hundreds of brands. Dynamic Yield allowed McDonald’s drive-throughs to quickly personalize menu boards based on a customer’s order and other factors. Company executives claimed the personalization technology boosted the average check, but in 2022, McDonald’s sold Dynamic Yield to Mastercard. Industry observers opined that the sale meant large retailers prefer to get AI services from specialist companies rather than supporting AI in-house themselves.

Lowe's icon.


Using Nvidia’s AI-based omniverse technology, Lowe’s built a digital twin deployment that allows the store’s retail assistants to quickly see and interact with the retailer’s digital data. The goal is to streamline operations and improve customer service. The AI system will also power a virtual 3-D product catalog.

Bloomreach icon.


A prime example of an AI vendor for the retail sector, Bloomreach’s solutions include Discovery, an AI-driven search and merchandising solution; and Engagement, a consumer data platform. This type of stand-alone AI vendor serving an industry vertical is likely to flourish because many large companies are not equipped to develop AI tool sets themselves.

Accenture icon.


Consulting giant Accenture’s ai.RETAIL solution enables retailers to use AI to turn data  — which retailers have reams of — into action that boosts the bottom line. The initiative includes dynamic merchandising, providing more real-time actionable data to store clerks, and driving predictive insights to stay ahead of retail trends.

Standard AI icon.

Standard AI

Clearly the wave of the future, Standard AI is an AI platform that allows customers browsing in stores to select and buy their item choices without the delay of paying a cashier. The strategy is “autonomous retail,” in which retail locations are retrofitted with AI technology to streamline the shopping experience.

Lucyd icon.


So you’ve been waiting for the first ChatGPT-enabled eyewear? Wait no more: Lucyd, a retailer of “smart” eyewear under the Eddie Bauer and Nautica brand names, has unveiled a smartphone app that allows you to speak to your glasses and hear responses through tiny speakers. The “wearables” sector now has a niche called “hearables.”

Veesual icon.


Veesual is an AI-powered virtual-try-on app that allows users to customize their outfits, virtual models, and the digital dressing room where they try on clothing. The tool uses deep learning so clothing images look realistic and maintain their definition when merged with human model images. Additionally, Veesual’s CX-focused approach to AI pays attention to finding and showing customers the best sizes for their needs.

Companion icon.


An AI-powered companion for your dog, Companion’s box (about the height of an average dog) uses machine vision and machine learning to interact with your pet in real time. The device can even dispense treats, which should help with any behavioral training goals. The company also plans on an AI companion for cats; given feline insouciance, the training modules might not be so well received.

Also see: ChatGPT: Understanding the ChatGPT ChatBot

AI Industry Organizations

These industry organizations for the AI sector play a number of crucial roles. First and foremost, they advocate for the regulation of artificial intelligence. This is an enormously important focus, given AI’s exponential growth will affect business and culture. To what extent can we as a society impose guidelines on AI’s growth, which has thus far been driven by pure profit?

These groups also lobby for greater diversity in AI, which is essential. We’ve already seen that AI systems embody legacy bias; this must be corrected more proactively to create inclusive systems. Additionally, these AI organizations support cross-vendor development of AI to promote the overall advancement of the technology.

Association for the Advancement of AI icon

Association for the Advancement of AI

Founded in 1979, the AAAI is an international scientific group focused on promoting responsible AI use, improving AI education, and offering guidance about the future of AI. It gives out a number of industry awards, including the AAAI Squirrel AI Award for Artificial Intelligence for the Benefit of Humanity, which provides $1 million to promote AI’s efforts to protect and enhance human life.

AI4Diversity icon.


This nonprofit’s motto is “Leveraging AI, education, and community-driven solutions to empower diversity and inclusion.” AI4Diversity was founded by Steve Nouri, a social media influencer and AI evangelist at Wand. Given that AI platforms have been found to perpetuate the bias of their creators, this focus on diversity and inclusion is essential.

AI Infrastructure Alliance icon.

AI Infrastructure Alliance

Supported by a group of major enterprise vendors that includes Hewlett Packard Enterprise, and sponsored by the likes of Nvidia, the AI Infrastructure Alliance “aims to foster collaboration and interoperability between leading MLOps tools to allow a CS [canonical stack] to form more quickly and effectively.” The organization supports open-source and open-core software so users aren’t locked into narrow proprietary solutions.

Partnership on AI icon.

Partnership on AI

Founded by a consortium of tech giants — Google, Meta, Amazon, IBM, and Microsoft — Partnership on AI is a nonprofit with a mission to research best practices for AI systems. It works to “bring together diverse voices from across the AI community.” Partnership on AI includes more than 100 partners from academia and business.

Black in AI icon.

Black in AI

Founded in 2017, Black in AI is a technology research and advocacy group dedicated to increasing the presence of black tech professionals in artificial intelligence. Black in AI notes that “representation matters,” and that AI algorithms are trained on data that reflects a legacy of discrimination, so promoting black voices in AI development is crucial to the technology’s growth.

Machine Intelligence Research Institute icon

Machine Intelligence Research Institute (MIRI)

Originally known as the Singularity Institute for Artificial Intelligence, MIRI supports research to “ensure that smarter-than-human artificial intelligence has a positive impact.” Among the recent cautionary articles that MIRI has posted: Pausing AI Developments Isn’t Enough. We Need to Shut it All Down.

AI Now Institute icon.

AI Now Institute

AI Now Institute creates policy research to address the concentration of power in the tech world. Their 2023 report Confronting Tech Power notes that “there is no AI with big tech,” and that “a handful of private actors have accrued power and resources that rival nation-states while developing and evangelizing artificial intelligence as critical social infrastructure.”

The Alan Turing Institute icon.

The Alan Turing Institute

Funded by the UK government, The Alan Turing Institute produces research that addresses crucial issues in artificial intelligence, society, and the economy and collaborates with businesses and public groups to use the research to deal with pressing concerns. That this group is government-funded raises a major question: Will more governments around the world step up to fund groups that prompt the AI sector to work for greater social good?

The Rockefeller Foundation icon.

The Rockefeller Foundation

While AI is only one of many focuses for this famed nonprofit, the Rockefeller Foundation is quite active in the AI sector; one of its core focuses is the responsible governance of AI. They issued a report called AI+1: Shaping Our Integrated Future, which is based on conclusions from a diverse group of experts who seek to deploy machine learning for positive social impact. Additionally, the foundation makes grants, including donating $500,000 to Black in AI.
eWeek video: Rockefeller Foundation’s Zia Khan on AI and Ethics

For more information, also see: History of AI

Bottom Line: AI Companies

This list of AI companies is, admittedly, a partial portrait. In truth, it’s a blurry snapshot of something whizzing by too fast to completely capture. The generative AI landscape in particular changes daily, sometimes hourly it seems. Each morning we’re greeted with a slew of headlines announcing new investments, fresh solutions, and surprising innovations that leap forward at a breakneck pace.

The progress of artificial intelligence won’t be linear because the nature of AI technology is inherently exponential. Today’s hyper-sophisticated algorithms, devouring more and more data, learn faster as they learn. It’s this exponential pace of growth in artificial intelligence that makes the technology’s impact so impossible to predict — which, again, means this list of leading AI companies will shift quickly and without notice.

As investment pours in, the underlying technologies that fuel artificial intelligence are each seeing their own rocket blasts of innovation. Machine learning, deep learning, neural networks, generative AI — legions of researchers and developers are creating a wild profusion of generative AI use cases. This is happening in facilities across the globe, in academia and business, by both good folks and decidedly not good. The race is on.

In past decades of the tech business, the incumbent market leaders would watch an innovative challenger, sense the threat, then acquire them and start selling the advanced tools as their own. In the AI sector, the closest thing we have to incumbents are the cloud leaders: AWS, Microsoft, and Google. Certainly, they’ve invested in exciting innovators: Google bought DeepMind and Microsoft has embraced OpenAI.

But the challenge facing these giants is that the world outside the castle walls is moving far too fast to control. Oh, they can distribute fantastical dollops of money all around, but even the most deep-pocketed cloud giant can’t afford to snap up all the innovative AI challengers. From AI in healthcare to AI in education, and all those niche companies chasing all those expanding AI use cases: no incumbent can dominate it all.

In sum, the life cycle for these AI companies is not so much digital transformation as digital revolution. Please check back to see the next version of this list — it’s very much a living document.

AI Companies market growth

Artificial intelligence market size worldwide, forecast to 2030. Source: Statistica.

The post Top 150+ Artificial Intelligence (AI) Companies 2024 appeared first on eWEEK.

Qumulo’s New Scale Anywhere Platform Aims to Modernize Data Storage https://www.eweek.com/big-data-and-analytics/qumulo-introduces-new-scale-anywhere-platform/ Fri, 22 Dec 2023 19:44:17 +0000 https://www.eweek.com/?p=223553 Cloud-native Qumulo unifies and simplifies access to data across the cloud spectrum

The post Qumulo’s New Scale Anywhere Platform Aims to Modernize Data Storage appeared first on eWEEK.

Seattle-based Qumulo, which describes itself as “the simple way to manage exabyte-scale data anywhere,” recently announced a new version of its Scale Anywhere platform.

The solution, which can run on commodity hardware or in the public cloud, seeks to help enterprises vexed by unstructured data. The company says that Scale Anywhere uses a unified approach to improve efficiency, security, and business agility.

In a briefing with ZK Research, Qumulo CTO Kiran Bhageshpur gave me some background on the platform. “We look at this as being the third era of unstructured data,” he told me. “The first era was NetApp with scale-up, dual controller architectures, and millions of files. It was really a sort of analysis box, if you will. The second era was Isilon, then EMC Isilon, now Dell EMC Isilon, which is scale-out storage, hardware appliances, on-premises, lots of them together to form large single volumes.”

Cloud-Based Qumulo Competes with Legacy Systems

Kiran said that Qumulo started in the cloud computing era, looked at the world, and realized it was no longer the scale-up or scale-out era.

“This is the scale-anywhere era of large-scale data,” he said. “It’s not only lots of data in the enterprise data center—there is incredible growth in the cloud and out at the edge. And Qumulo, with a pure software solution, can now present a solution for all of this data—cloud, on-premises, and the edge in one consistent way.”

Qumulo says that Scale Anywhere introduces a way for enterprises to use on-premises storage in a similar way to cloud storage.

The company jointly developed Azure Native Qumulo (ANQ) with Microsoft. This cloud-native enterprise file system helps eliminate the tradeoffs that often come with balancing scale, economics, and performance.

Qumulo is trumpeting a number of advantages to the approach, including:

  • Affordability: Qumulo says that ANQ is about 80% cheaper than competitive offerings and compares well to the costs of traditional on-premises storage.
  • Elasticity: Qumulo says that ANQ separates the scalability of capacity and performance so they can operate independently.
  • Cloud configurable: Qumulo says enterprises can use the Azure service portal to configure and deploy ANQ quickly.
  • Data services: Qumulo says that ANQ provides several data services, including quotas, snapshots, multi-protocol access, enterprise security integrations, and real-time data analytics.

The company also announced Qumulo Global Namespace (Q-GNS), which acts as a unified data plane for unstructured data.

“This is the core feature of the underlying Qumulo file system, and it allows the customer to access remote data on a remote Qumulo cluster as if it were local,” Kiran told me. “Think of two, three, or four Qumulo clusters talking to each other. You can connect to the local one. And as long as it’s configured correctly, you can access data on a Qumulo cluster in the cloud or on-premises halfway across the world, and it feels as though it were local.”

In the announcement, JD Whitlock, CIO of Dayton Children’s Hospital, said that his hospital uses Q-GNS.

“We are rapidly adopting cloud to store our long-term radiology images while keeping new images on-premises,” Whitlock said. “Qumulo’s Global Namespace makes it easy to bring our file-based workloads to the cloud without refactoring any applications.”

Also see: Top Cloud Service Providers and Companies

Bottom Line: Storage for the Cloud Era

Legacy storage vendors like Dell EMC view data storage as an entitlement and haven’t delivered innovation in years. Many believe storage to be a commodity with little room for new features and functions, but that’s not true. The announcement by Qumulo modernizes storage for the cloud era. The company has a lot of work ahead of it, but the approach is innovative and might just make a dent in the defenses of the legacy players.

Read next: Top Digital Transformation Companies

The post Qumulo’s New Scale Anywhere Platform Aims to Modernize Data Storage appeared first on eWEEK.

The Ultimate Guide to Prompt Engineering https://www.eweek.com/artificial-intelligence/guide-to-prompt-engineering/ Thu, 21 Dec 2023 16:59:40 +0000 https://www.eweek.com/?p=223556 Learn the basics of prompt engineering and how it can help you optimize your storage system. Discover the best practices and get started today.

The post The Ultimate Guide to Prompt Engineering appeared first on eWEEK.

Prompt engineering is the art and process of instructing a generative AI software program to produce the output you want, accurately and efficiently.

This task is far from easy, due to the enormous range of output created by the advanced algorithms included in generative AI programs. Additionally, the vast data repositories in large language models – data stores that are growing daily – enables generative AI programs to offer an enormous range of outputs.

Consequently, effective prompt engineering requires an overall understanding of prompt engineering, knowing the key tips for a well-crafted prompt, and being aware of key use cases for an effective AI prompt.

Understanding Prompt Engineering

Generative artificial intelligence has become ubiquitous, with a wide array of generative apps and tools for both professional and recreational use. This is largely due to the success and popularity garnered by ChatGPT, Google Bard or any AI chatbot that leverages a large language model.

Users have realized that getting the best answer from a generative AI app can be challenging. One common misunderstanding: users enter casually-worded prompts as if the generative AI program were a fast Google search. This won’t produce optimal results.

Adding to the challenge, today’s generative AI programs span all types of multimedia:

  • Image generators, like Open AI Dall-E and Midjourney ,which are ideal for generating 3D assets and AI art.
  • Code generators, like Copilot, which are used for generating code snippets and assisting developers in coding tasks.
  • Speech synthesis models, like DeepMind’s WaveNet, are used to generate realistic human-like speech.

In sum, there’s growing realization about using generative AI tools: your output is only as good as your prompt. And your prompt must be constructed to include elements based on not just your core concept, but also what type of media (photo, video) you are using to build your content.

A generative AI program is powerful, but it cannot guess your intent – you need to tell it exactly what you want.

How to Engineer a Well-Crafted Prompt: 7 Tips

A well-engineered prompt encompasses skills and techniques that enable you to understand the capabilities and limitations of a model, and then tailor your prompt to optimize the quality and relevance of the generated output.

A carefully crafted prompt should include several core elements, including:

1) Provide Context

This is the big one. Using proper context in your prompt means including all relevant background information, from crucial details and any necessary instructions. Are you creating a prompt in the context of healthcare, or finance, or software coding? Include that context. Similarly, is your prompt designed to output historic material or a forward-looking response? Include all important time-sensitive context in your prompt’s verbiage.

For example, you may want to generate information about the scientific process of photosynthesis. If you’re used to Google, you might simply type the prompt “photosynthesis.”

With a vague, one-word prompt, you would still get some results:

ChatGPT example prompt

However, for a fuller, more detailed sense of the biological underpinnings of photosynthesis, look at what results from a prompt with fuller and deeper context:

ChatGPT example prompt

2) Be Specific 

Clearly define the task or question you want the AI model to address. Be highly specific about what information or output you are expecting: add more detail than you think you might need. For instance, specify if you are seeking data or any calculations in your output, and include the full array of project-related data that will enable the LLM to find the one needle in a haystack you’re searching for.

3) Add All Constraints

If there are any limitations or constraints that the AI model needs to consider, such as word count, time frame or specific criteria, make them explicit in the prompt. Realize that all generative AI software have access to an ocean of data; without very tight constraints built into your prompt, you’re asking for more output than you actually want.

4) Use System-Related Instructions

Many generative AI models have a “system” – often thought of as a “user persona” – that specific prompt instructions can influence. Use these instructions to guide the model’s behavior or style of output.

For instance, if you’re really looking for a “dad” joke, then go ahead and spell that out clearly in your prompt: “tell me a joke in the style of a dad joke.” After using this system-related instruction, you can easily produce dozens of dad jokes.

5) Experiment with Temperature and Top-K and Top-P

Adjusting parameters such as temperature, top-k, and/or top-p can control the randomness or creativity of the model’s output. As you craft your prompt, experiment with these parameters to fine-tune the generated output based on your preferences.

Temperature Value

​The temperature value refers to the randomness or creativity of the generated responses. A higher temperature value, such as 0.8 or 1.0, makes the output more random and unpredictable, while a lower temperature value, such as 0.2 or 0.5, makes the output more focused and deterministic. Most likely, you’ll want to create a more formal response for a business project and a more random approach for a recreational project.

Temperature values and what they mean:

Temperature value Result
0.0 to 0.3 Formal and non-creative
0.3 to 0.7 Balanced tone
0.7 to 1.0 Creative and semi-random tone

Top K Value

The top k value is the parameter that determines the number of next tokens most likely to be considered during text generation. Put simply, a higher top k value allows for a more diverse and creative output as more alternative tokens are considered. Conversely, a lower top k value results in more conservative and predictable responses.

Top K values and what they mean:

Top-k range Result
1 to 5 More focused and concise responses.
5 to 10 Balanced responses with a mix of common and nuanced information. Improved coherence.
20+ More diverse and detailed responses. It may include more creative elements but could be less focused.

Top P Value

Top p value is an alternative method to control the diversity of generated responses during text generation.

Instead of considering a fixed number of most likely tokens like top k, top p sampling involves considering the tokens until their cumulative probability reaches a certain threshold or percentage. This threshold is dynamic and can vary with each token generation step.

For example, if the top-p value is set to 0.8, the model will consider tokens until the cumulative probability of those tokens reaches 80%.

Top P values and what they mean:

Top-p range Result
0.1 to 0.3 More focused responses with a narrow set of possibilities. Higher probability for common and safe responses.
0.3 to 0.6 Balanced responses with a mix of common and diverse information. Offers more flexibility in generating creative and nuanced content.
0.7 to 1.0 Greater diversity in responses. It may include more unexpected or creative elements but be less focused or coherent.

6) Test and Iterate Your Prompt

You will likely try various prompts to more efficiently create the output you desire. This is not only a productive approach, it’s one of the most effective possible strategies for improving your prompt engineering.

So don’t hesitate to experiment – again and again – with different prompts, and then evaluate the generated output  to refine and improve your prompt engineering. As prompt experts have learned, you can’t become a top-level prompt engineer without this extensive iteration.

Important point: even if you think you’re getting the output you need, don’t hesitate to dig deeper and try again. There’s no downside to continued attempts.

7) Understand Model Limitations

This is a hard fact for many prompt engineers to accept: a given generative AI model may include an excellent large language model, but that model cannot answer every question. For instance, a large language model that was built to serve the healthcare market may not be able to handle the most upper level finance-related queries that an MBA can type into it.

So it’s essential to engineer the best prompts within the limitations and capabilities of the specific AI model you are using. This will help you design prompts that work well with the model’s strengths and mitigate potential pitfalls of limited outputs.

Use Cases for Prompt Engineering

If you’re wondering about the possible use cases for prompt engineering, realize that the number is far from fixed – the number of potential use cases grows constantly.

As noted below, they currently range from creating blog posts to full-length content summarization. As the generative AI tools grow – and as the field of prompt engineering advances – the number of use cases expands across finance, manufacturing, healthcare and many more fields.

A sampling of current use cases for prompt engineering:

Application/Use cases Description
Content generation Create blog posts, articles and creative writing
Code generation Code generation and debugging assistance
Language translation Prompt the model to translate text between different languages
Conversational agents Develop conversational agents like chatbots
Idea generation Generate ideas for various projects, businesses or creative endeavors
Content summarization Summarize lengthy documents, articles or reports
Simulation and scenario exploration What-If Analysis – explore hypothetical scenarios and their potential outcomes
Entertainment Generate jokes, puns, or humorous content
Persona creation Generate personas for storytelling, gaming, or role-playing scenarios
General question answering Pose questions to the model to gather information on a wide range of topics

5 Top Prompt Engineering Tools and Software

  • Azure Prompt Flow: Powered by LLMs, the Azure Prompt Flow is currently one of the most sophisticated and feature-rich prompt engineering tools. It enables you to create executable flows that connect LLMs, prompts and Python tools through a visualized graph.
  • Agenta: An open-source tool designed for building LLM applications. It offers resources for experimentation, prompt engineering, evaluation and monitoring, allowing you to develop and deploy LLM apps.
  • Helicone.ai: This is a Y Combinator-backed open-source platform for AI observability. It provides monitoring, logging and tracing for your LLM applications, which can help write prompts.
  • LLMStudio: Developed by TensorOps, LLMStudio is an open-source tool designed to streamline your interactions with advanced language models such as OpenAI, VertexAI, Bedrock and even Google’s PaLM 2.
  • LangChain: This is a framework for developing applications powered by language models. It includes several modules such as LangChain libraries, LangChain templates, LangServe and LangSmith, which is a platform designed to simplify debugging, testing, evaluating and monitoring large language model applications.

Bottom Line: Effective Prompt Engineering

Prompt engineering is an art and science that requires skills, reasoning and creativity. It is a dynamic skill that merges the precision of science with the artistry of advanced language. When mastered and used effectively, it becomes a powerful tool for harnessing the capabilities of generative AI software.

Of course prompt engineering as a field is brand new – the term hardly existed even as recently as the end of 2022. So be aware that the strategies and best practices for prompt engineering will change rapidly in the months and years ahead. Also realize that prompt engineering is a valuable skill in the AI market; if you’re a top prompt engineering, there will likely be an open job for you.

To gain a fuller understanding of generative AI, read our guide: What is Generative AI? 

The post The Ultimate Guide to Prompt Engineering appeared first on eWEEK.

What Are AI Hallucinations and How To Stop Them https://www.eweek.com/artificial-intelligence/ai-hallucinations/ Wed, 20 Dec 2023 23:06:04 +0000 https://www.eweek.com/?p=223506 AI hallucinations are divergent outputs that may be inaccurate or offensive. Learn what they are, how they work, and what you can do to mitigate them here.

The post What Are AI Hallucinations and How To Stop Them appeared first on eWEEK.

An AI hallucination is an instance in which an AI model produces a wholly unexpected output; it may be negative and offensive, wildly inaccurate, humorous, or simply creative and unusual.

AI hallucinations are not inherently harmful — in fact, in certain creative and artistic contexts, they can be incredibly interesting — but they are statements that aren’t grounded in fact or logic and should be treated as such.

Some AI hallucinations are easy to spot, while others may be more subtle and go undetected. If users fail to identify an AI hallucination when it occurs and so pass off this output as fact, it can lead to a range of issues. In any case, if your work relies on an generative AI program, it’s essential to prevent AI hallucinations – see our 7 methods below.

Understanding AI Hallucinations

AI models — especially generative AI models and LLMs — have grown in the complexity of the data and queries they can handle, generating intelligent responses, imagery, audio, and other outputs that typically align with the users’ requests and expectations.

However, artificial intelligence models aren’t foolproof; because of the massive amounts of training data, complicated algorithms, and other less-than-transparent factors that go into preparing these models for the market, many AI platforms have run into issues where they “hallucinate” and deliver incorrect information.

For instance, in this example, ChatGPT attempts to translate “Lojban” – which is a fake language – into English, with semi-hilarious results:

Example of an AI hallucination

However, an AI hallucination is not always a negative thing, as long as you know what you’re dealing with; certain AI hallucinations, especially with audio and visuals, can lead to interesting elements for creative brainstorming and project development.

7 Methods for Detecting and Preventing AI Hallucinations

The following techniques are designed for AI model developers and vendors as well as for organizations that are deploying AI as part of their business operations, but not necessarily developing their own models.

Please note: These techniques are not focused on the actual inputs users may be putting into the consumer-facing sides of these models. But if you’re a consumer looking to optimize AI outcomes and avoid hallucinatory responses, here are a few quick tips:

  • Submit clear and focused prompts that stick to one core topic; you can always follow up with a separate query if you have multiple parts to your question.
  • Ensure you provide enough contextual information for the AI to provide an accurate and well-rounded response.
  • Avoid acronyms, slang, and other confusing language in your prompts.
  • Verify AI outputs with your own research.
  • Pay attention; in many cases, AI hallucinations give themselves away with odd or outlandish language or images.

Now, here’s a closer look at how tech leaders and organizations can prevent, detect, and mitigate AI hallucinations in the AI models they develop and manage:

1) Clean and Prepare Training Data for Better Outcomes

Appropriately cleansing and preparing your training data for AI model development and fine-tuning is one of the best steps you can take toward avoiding AI hallucinations.

A thorough data preparation process improves the quality of the data you’re using and gives you the time and testing space to recognize and eliminate issues in the dataset, including certain biases that could feed into hallucinations.

Data preprocessing, normalization, anomaly detection, and other big data preparation work should be completed from the outset and in some form each time you update training data or retrain your model. For retraining in particular, going through data preparation again ensures that the model has not learned or retained any behavior that will feed back into the training data and lead to deeper problems in the future.

2) Design Models With Interpretability and Explainability

The larger AI models that so many enterprises are moving toward have massive capabilities but can also become so dense with information and training that even their developers struggle to interpret and explain what these models are doing.

Issues with interpretability and explainability become most apparent when models begin to produce hallucinatory or inaccurate information. In this case, model developers aren’t always sure what’s causing the problem or how they can fix it, which can lead to frustration within the company and among end users.

To remove some of this doubt and confusion from the beginning, plan to design AI models with interpretability and explainability, incorporating features that focus on these two priorities into your blueprint design. While building your own models, document your processes, maintain transparency among key stakeholders, and select an architecture format that is easy to interpret and explain, no matter how data and user expectations grow.

One type of architecture that works well for interpretability, explainability, and overall accuracy is an ensemble model; this type of AI/ML approach pulls predicted outcomes from multiple models and aggregates them for more accurate, well-rounded, and transparent outputs.

3) Test Models and Training Data for Performance Issues

Before you deploy an AI model and even after the fact, your team should spend significant time testing both the AI models and any training data or algorithms for performance issues that may arise in real-world scenarios.

Comprehensive testing should cover not only more common queries and input formats but also edge cases and complex queries. Testing your AI on how it responds to a wide range of possible inputs predicts how the model will perform for different use cases. It also gives your team the chance to improve data and model architecture before end users become frustrated with inaccurate or hallucinatory results.

If the AI model you’re working with can accept data in different formats, be sure to test it both with alphanumeric and audio or visual data inputs. Also, consider completing adversarial testing to intentionally try to mislead the model and determine if it falls for the bait. Many of these tests can be automated with the right tools in place.

4) Incorporate Human Quality Assurance Management

Several data quality, AI management, and model monitoring tools can assist your organization in maintaining high-quality AI models that deliver the best possible outputs. However, these tools aren’t always the best for detecting more obscure or subtle AI hallucinations; in these cases and many others, it’s a good idea to include a team of humans who can assist with AI quality assurance management.

Using a human-in-the-loop review format can help to catch oddities that machines may miss and also give your AI developers real-world recommendations for how improvements should be made. The individuals who handle this type of work should have a healthy balance of AI/technology skills and experience, customer service experience, and perhaps even compliance experience. This blended background will give them the knowledge they need to identify issues and create better outcomes for your end users.

5) Collect User Feedback Regularly

Especially once an AI model is already in operation, the users themselves are your best source of information when it comes to AI hallucinations and other performance aberrations. If appropriate feedback channels are put in place, users can inform model developers and AI vendors of real scenarios where the model’s outputs went amiss.

With this specific knowledge, developers can identify both one-off outcomes and trending errors, and, from there, they can use this knowledge to improve the model’s training data and responses to similar queries in future iterations of the platform.

6) Partner With Ethical and Transparent Vendors

Whether you’re an AI developer or an enterprise that uses AI technology, it’s important to partner with other ethical vendors that emphasize transparent and compliant data collection, model training, model design, and model deployment practices.

This will ensure you know how the models you use are trained and what safeguards are in place to protect user data and prevent hallucinatory outcomes. Ideally, you’ll want to work with vendors that can clearly articulate the work they’re doing to achieve ethical outcomes and produce products that balance accuracy with scalability.

To gain a deeper understanding of AI ethical issues, read our guide: Generative AI Ethics: Concerns and Solutions

7) Monitor and Update Your Models Over Time

AI models work best when they are continuously updated and improved. These improvements should be made based on user feedback, your team’s research, trends in the greater industry, and any performance data your quality management and monitoring tools collect. Regularly monitoring AI model performance from all these angles and committing to improving models based on these analytics can help you avoid previous hallucination scenarios and other performance problems in the future.

IBM's Watson OpenScale is an open platform that helps users govern AI and manage fairness, drift, and other quality issues.
IBM’s Watson OpenScale is an open platform that helps users govern AI and manage fairness, drift, and other quality issues. Source: IBM.

How and Why Do AI Hallucinations Occur?

It’s not always clear how and why AI hallucinations occur, which is part of why they have become such a problem. Users aren’t always able to identify hallucinations when they happen and AI developers often can’t determine what anomaly, training issue, or other factor may have led to such an outcome.

The algorithms on which modern neural networks and larger AI models are trained are highly complex and designed to mimic the human brain. This gives them the ability to handle more complex and multifaceted user requests, but it also gives them a level of independence and seeming autonomy that makes it more difficult to understand how they arrive at certain decisions.

While it does not appear that AI hallucinations can be eliminated at this time, especially in more intricate AI models, these are a few of the most common issues that contribute to AI hallucinations:

  • Incomplete training data.
  • Biased training data.
  • Overfitting and lack of context.
  • Unclear or inappropriately sized model parameters.
  • Unclear prompts.

Issues That May Arise From AI Hallucinations

AI hallucinations can lead to a number of different problems for your organization, its data, and its customers. These are just a handful of the issues that may arise based on hallucinatory outputs:

  • Inaccurate decision-making and diagnostics: AI instances may confidently make an inaccurate statement of fact that leads healthcare workers, insurance providers, and other professionals to make inaccurate decisions or diagnoses that negatively impact other people and/or their reputations. For example, based on a query it receives about a patient’s blood glucose levels, an AI model may diagnose a patient with diabetes when their blood work does not indicate this health problem exists.
  • Discriminatory, offensive, harmful, or otherwise outlandish outputs: Whether it’s the result of biased training data or the rationale is completely obscure, an AI model may suddenly begin to generate harmfully stereotypical, rude, or even threatening outputs. While these kinds of outlandish outputs are typically easy to detect, they can lead to a range of issues, including offending the end user.
  • Unreliable data for analytics and other business decisions: AI models aren’t always perfect with numbers, but instead of stating when they are unable to come to the correct answer, some AI models have hallucinated and produced inaccurate data results. If business users are not careful, they may unknowingly rely on this inaccurate business analytics data when making important decisions.
  • Ethical and legal concerns: AI hallucinations may expose private data or other sensitive information that can lead to cybersecurity and legal issues. Additionally, offensive or discriminatory statements may lead to ethical dilemmas for the organization that hosts this AI platform.
  • Misinformation related to global news and current events: When users work with AI platforms to fact-check, especially for real-time news and ongoing current events, depending on how the question is phrased and how recent and comprehensive the AI model’s training is, the model may confidently produce misinformation that the user may spread without realizing its inaccuracies.
  • Poor user experience: If an AI model regularly produces offensive, incomplete, inaccurate, or otherwise confusing content, users will likely become frustrated and choose to stop using the model and/or switch to a competitor. This can alienate your core audience and limit opportunities for building a larger audience of users.

Read next: 50 Generative AI Startups to Watch

Bottom Line: Preventing AI Hallucinations When Using Large-Scale AI Models

The biggest AI innovators recognize that AI hallucinations create real problems and are taking major steps to counteract hallucinations and misinformation, but AI models continue to produce hallucinatory content on occasion.

Whether you’re an AI developer or an enterprise user, it’s important to recognize that these hallucinations are happening, but, fortunately, there are steps you can take to better identify hallucinations and correct for the negative outcomes that accompany them. It requires the right combination of comprehensive training and testing, monitoring and quality management tools, well-trained internal teams, and a process that emphasizes continual feedback loops and improvement. With this strategy in place, your team can better address and mitigate AI hallucinations before they lead to cybersecurity, compliance, and reputation issues for the organization.

For more information about governing your AI deployment, read our guide: AI Policy and Governance: What You Need to Know

The post What Are AI Hallucinations and How To Stop Them appeared first on eWEEK.