The Ouroboros of Technology: What If AI Is Eating the Old World to Create a New One?

The ouroboros: the old form consumed to become the material of the new one.

There's an image I keep coming back to when I think about AI: the ouroboros, the ancient symbol of a serpent eating its own tail — usually read as a cycle of creation and destruction, something consuming itself to become something new. That feels close to what's happening in technology right now. For decades the industry built increasingly powerful layers of abstraction — machine language to assembly, assembly to high-level languages, servers to virtual machines to containers, infrastructure to APIs, software to platforms — each one taking something that required specialized knowledge and making it accessible to more people. Then we built AI, and it appears to be consuming the very technical skills that produced it. That's understandably frightening.

The fear is not irrational

There's real anxiety in the industry right now, and most of it centers on jobs. Developers watch systems generate working code from a paragraph of English. Designers see models producing interfaces. Writers see models producing prose. Support engineers see agents diagnosing systems. It's easy to look at that trajectory and see only subtraction — fewer programmers, fewer designers, fewer analysts, fewer people needed to turn an idea into a product. Some of that will probably happen; it would be foolish to pretend otherwise.

But there's another possibility that's harder to see because we're standing in the middle of the transition: what if we're focused too much on the jobs AI is consuming, and not enough on the capabilities it's distributing?

What happens when almost anyone can build?

Until recently, having an idea and building a working product were two entirely different things. A brilliant idea went nowhere without developers, infrastructure, design, database expertise, deployment pipelines, money — above all, access to specialized knowledge. That created an enormous gap between imagining something and making something.

AI is starting to close it. A person can describe an application in plain language and watch pieces of it appear: the database schema, the API, the interface, the deployment configuration — then run it, find what's wrong, explain what they don't like, and iterate. That doesn't mean expertise stops mattering; someone who actually understands architecture, security, or infrastructure will guide these systems far more effectively than someone who doesn't. But the minimum expertise required to begin creating is dropping fast, and that may turn out to be one of the more important consequences of AI.

We may all become product managers

In a sense, AI turns the user into something resembling a product manager: you describe what you want, define constraints, look at what comes back, decide what's wrong, refine the requirements, and make judgment calls about what matters. Those were traditionally higher-level activities performed by people who could rely on an engineering organization underneath them. AI is starting to provide something that looks — imperfectly, inconsistently, sometimes maddeningly — like that organization.

The interesting consequence is that the ability to create software may no longer belong primarily to people whose profession is creating software. A teacher might build a tool for her own classroom. A musician might build software for practicing an instrument. A farmer might build something for a problem unique to his land. A small-business owner might automate a workflow no software company would ever find economically interesting. Historically, most of those ideas simply died — not because they were bad, but because the cost of turning them into reality was too high. AI changes that math.

The startup founder without the startup

This also changes what it means to be an entrepreneur. For most of the software era, starting a company meant assembling a small institution before you could seriously test an idea — technical cofounders, capital to pay them, months of development before you knew whether anyone wanted what you were building. Increasingly, a person may get much farther alone. Not necessarily all the way to a mature company — scaling, security, reliability, sales, and good judgment remain stubbornly real problems — but perhaps far enough to answer the most important early question: does this idea deserve to exist?

That's significant. It means experimentation gets cheaper, and cheaper experimentation means more experiments. Most will fail, some will be pointless — but hidden among them will be things that could never have existed when creation required permission from capital, employers, or technical gatekeepers.

The creative impulse gets set free

This may end up being bigger than software. People are astonishingly good at having ideas and remarkably bad at bringing many of them into the world — books never written, businesses never started, tools never built, songs never finished. Some of that is ordinary procrastination. Some of it is friction: too much distance between the idea in your head and the skills required to make it real. Generative AI is compressing that distance.

The technology may commoditize execution while making distinctly human qualities more important — which isn't necessarily the dystopian outcome it's often framed as.

But we can't see far enough ahead

None of this means the transition will be painless. Technological revolutions rarely distribute their benefits neatly — people can lose livelihoods long before new opportunities emerge, institutions adapt slowly, and gains can concentrate in a small number of hands. It would be naive to declare that everything will simply work itself out. But it may be equally naive to assume today's job categories are the final form of human economic life.

We tend to imagine the future by subtracting from the present: today's economy, minus programmers; today's companies, minus designers. That's probably the wrong model. Transformative technologies don't just remove pieces of the existing world — they change what becomes possible. Someone standing at the start of the internet could reasonably have worried about what email would do to postal workers, or what online commerce would do to retail stores. Those were legitimate questions. But from that vantage point it would have been almost impossible to imagine YouTube creators, cloud architects, app developers, cybersecurity engineers, or the millions of other forms of work that showed up later. Destruction is visible immediately. Creation usually isn't.

The serpent eating its tail

Which brings me back to the ouroboros. There's something almost paradoxical in what the technology industry has done: generations of engineers accumulated knowledge about how to instruct computers, then used that knowledge to build systems capable of absorbing enormous amounts of human knowledge — including knowledge about how to instruct computers. Now those systems are making some of that specialized knowledge accessible through ordinary language. The machinery of abstraction has reached the people who built the abstractions.

The serpent has reached its tail.

But the ouroboros doesn't traditionally represent simple self-destruction — it represents renewal. The old thing is consumed and becomes the material the next thing is made from. AI will eliminate certain kinds of work and dramatically change others, and there are serious open questions about who benefits and who gets displaced. Those questions deserve real weight.

But I suspect we're witnessing something larger than automation: the democratization of creation itself — a world where a lot more people can move from I have an idea to let me see if I can build it. The defining skill in that world may not be knowing how to build every piece of the machine. It may be knowing what machine ought to exist. That's a very different future than one where humans simply become obsolete — and from where we're standing, we probably can't see very far into it yet.