This was originally posted as a Medium article I wrote way back in early 2025.
I see a pervading idea in the AI hype-space that “AI” will replace programmers and those who build apps and software services. The idea is that these people are at risk of becoming obsolete because AI will replace them. The assumption seems to be that the company will be able to build apps and services much more cheaply and quickly by replacing these costly engineers with AI; continuing on as normal selling their app or service.
But if AI is actually going to advance to a level that replaces programmers, then it will inevitably replace companies that make software as well.
The problem with the idea is that if AI can accomplish what it once took an entire company to do with a team of engineers, then the company itself will be obsolete.
The “end stage” of AI progression brings with it not just the end of programmers, designers and those who build things for computers, but the end of the software company itself. If anyone can make use of AI to build just about any app or service they want, then the company-selling-a-service/app model is over. The way we interact with computers will fundamentally change. The average person will largely interact with computers through AI and quick niche interfaces and ephemeral apps that the AI can render for us on the fly. Each user will essentially have a collection of their own custom apps; not from any company, but built on the fly. The companies providing the ML models (OpenAI, Google, etc) will be part of a new class of utility services that will provide a new interface to the computer.
The other side of the coin for the idea that “software companies won’t be hiring programmers anymore” is, “software companies won’t exist anymore.” This old way of interfacing with computers will slowly go away. If every person has a little prompt box that can produce and generate just about any kind of interface or app, then stodgy old apps that you can’t modify or control and have to pay for or subscribe to will go the way of the buggy whip. When a user can type in, “make me a beautiful e-commerce website and interface to upload my products” and boom there it is, then we have no need for Shopify. If a user can type, “pull up an interface where I can read my email.” Then we don’t need Gmail. And it will be much “better” than Gmail since the user can add on to it infinitely. You wouldn’t have to wait for updates and new features to software. The only companies left will be those that provide data and raw apis to interface directly with AI.
There are so many software companies and startups desperately trying to ride the AI hype train and creating niche “AI powered” apps or integrate it into their existing apps. If the billions and billions of dollars invested and bold claims by company higher ups about AI’s future turn out to be accurate indicators of things to come, then these companies are just creating soon-to-be-legacy hybrid artifacts of the past. If it’s true that AI continues to get more and more powerful, then fairly soon the end user will be able to simply create interfaces and apps that will cater to their niche need in a matter of a few sentences. That’s the end result of AI; an entirely new way of interfacing with computers, not hybrid “AI powered apps.” These companies will have no way to provide unique value to the end user if the user has access to the thing that creates their value. It’s like thinking they’ll sell golden goose eggs to a world where everyone has a golden goose. A unique app or service is useless if any old user can generate one with a few sentences.
It’s as ridiculous as a company using combustion engines in their horse carriage factory. They think “we will increase productivity!”, “we can fire all these horse carriage engineers working on the assembly line, and sell more horse carriages!” I think you can see the irony there.
I don’t know the future of AI. I don’t personally think AI is anywhere near a level that will disrupt the industry like this, nor will it be any time soon or at all. I just wanted to point out the logical conclusion of what an endlessly progressing AI would mean for the software industry and for computing in general. On a logical level, given continued AI progression, the shiny new thing a lot of companies are so optimistic about will eventually obliterate their need for existence. AI will eat the entire industry. The end stage of AI progression in the computing world is that AI will be everything. It won’t help build things. It will be things. It will be the interface.
Either programmers will stick around and build things that continue to have value (at least a subset), or AI will take out the whole software industry (at least as it looks now); there’s not really any in-between. Which, for a programmer like me, is a comforting thought. If AI replaces the need for programmers, at least we’ll bring down the whole ship with us.