“Software is eating the world,” Marc Andreessen so famously observed in 2011. Yet now in 2021, it’s time to add a new phrase to his famous truism: “and artificial intelligence is eating software.”
Clearly, artificial intelligence will alter the software business at every level: how applications will function, how they’ll evolve, even how they’re sold. But likely the most revolutionary of these changes is how applications are created.
The AI technology driving this change is called various things, but the phrase “AI-Augmented software engineering” is as good as any. You’ll see it perched at the top of Gartner’s chart of emerging technologies:
What is AI-Augmented software development? In short: it’s a system of development tools and platforms with AI built in that enables exponentially faster and better app creation than “hand” coding or traditional dev tools.
Among other advantages, the AI-driven system does the grunt work of laying out code; it can even predict or suggest code frameworks.
AI Will Enable Citizen Developers
Perhaps most significant, AI enables less technically-inclined people to create or upgrade applications. Opening the gates of software creation to non-techies is a big disrupter – they vastly outnumber the slender cohort of skilled devs. While skilled developers will move faster with AI, the large pool of non-devs could provide a generational push to innovation.
Note that Gartner puts AI-Augmented software engineering at the very peak of “inflated expectations.” To be sure, this idea is (mostly) still a hope for the future, and has limits even in best case.
The problem is that writing software is like any upper-end intellectual endeavor: the judgment and nuance of the human mind are required for top work. Writing software is creative, as any good dev will tell you. Just as a song can’t be written by a computer (though “song-like music” can), a complex, new piece of software still can’t be coded by an AI system.
On the other hand, an AI system “learns” prodigiously, so it can suggest paths that might elude the most creative human. An AI-augmented software program takes in a torrent of data; it gains knowledge (or at least data) far faster and more comprehensively than humans. It can’t make the “leaps” of human developers, yet it can lay out patterns and fill in decision trees, or even predict future directions.
Low Code Platforms Begin to Incorporate AI-Augmented Software Development
AI-augmented software development is rising in tandem with the rapidly growing low code / no code market. A low code software platform offers an easy-to-understand visual interface that enables non-techies to build or tweak applications.
Major low code platforms are beginning to incorporate AI, notably Google’s AppSheet and Microsoft’s Power Platform. AppSheet uses natural language processing (NLP) to allow citizen developers to simply speak commands for the app’s development. Although in its infancy, this use of NLP is a futurist’s dream – creating software is as easy as talking to a computer.
AppSheet uses AI and ML to build predictive models into an application using the app’s own store of data. Remarkably, Google claims that this ML-intensive task requires no prior ML experience from the developer.
Similarly, Microsoft’s Power Platform includes Power Automate and Power BI modules to allow a non-tech developer to design and automate analytics systems into the application with relative ease. AI really is opening doors to an entirely new group of citizen developers.
This larger group of “developers” is needed. Adopting AI-Augmented software development is a necessity for companies to remain competitive. Developers are expensive and in short supply: US labor statistics indicate that there were 1.4 million computing science jobs that were unfilled in 2020. Companies routinely face challenges in hiring software developers.
Four Long Term Effects of AI-Augmentation
Clearly, AI-augmented software will dramatically shape the future: When writing software is as accessible as writing a detailed report, the pace of business will change in ways that aren’t fully predictable. Some reasonable assumptions:
- Data explosion It’s likely that most of the apps created with AI-assisted tools will mine, manipulate, or present data. Any capable staffer will be able to find new ways to use data for competitive advantage; your average sale rep will be altering apps to learn more about prospects. The end result is that data mining will grow even more parabolically than it is today.
- Security concerns: It’s reasonable to assume that lower level staffers won’t be able to code an application that will allow a major cyber attack; to prevent this, AI-augmented platforms will – we hope – have “guardrails” to block cybersecurity vulnerabilities by rookie devs. Yet with such vastly larger brigades of citizen developers, building so many intricate structures – getting more advanced as AI advances – it’s likely that we’ll see security holes.
- AI builds AI: In a boost to AI, AI-Augmented development platforms will be used to create more artificial intelligence capability. The process will fuel self-referential exponential growth: a tool that uses AI will create AI products, which in turn allows faster and more advanced building of AI-boosted applications. It is, perhaps, a dizzying prospect. Where the future takes us in this regard is hard to say. But when futurists talk about “the singularity” – when machines gain true independence – then this “AI builds AI” aspect clearly suggests it.
- Democratization of Tech: Certainly, the greatest effect of AI-augmented software is the democratization of software development and technology overall. Cloud computing allowed small companies (even startups) to rent a data center and so compete with far larger outfits. Similarly, AI-augmented software platforms will allow smaller companies to build out big time competitive infrastructure.
Bottom line: we will soon look back at today’s non-AI based software and wonder, how did we get anything done with these applications?