More on AI Programming

I posted previously on some experiences with AI programming, and I wanted to add to it a bit.

There's plenty of talk about AI (particularly in the programming space) eliminating jobs. I don't have any magic crystal ball, so who knows how it'll work out, but here's where I'm at presently.

I have this todo list of app ideas. When the inspiration hits me, I'll see if I can churn out a proof-of-concept on one of them, and maybe it'll even see the light of day. With AI though, that nights-and-weekends for a few months project becomes something I can proof-of-concept in a weekend, and get to production in weeks.

So, if AI can optimize my work, maybe we need fewer programmers? I offer this alternative position: the scope and significance of software will continue to increase.

Before compilers existed, the largest programs were incredibly tiny. It is notable that some impressive games were written entirely in assembler. Meanwhile, the magnitude of AAA games these days means that teams of 100s of engineers labor for years to produce a single game; often not a particularly good game, even. The bar has been raised in the gaming world.

Command line text editing was once the state of the art. This might not be the greatest example because I think Microsoft Word is truly horrific, but at the very least you can say with certainty that the notion of writing a document has taken on a whole new meaning, with software unimaginably more complex than what used to be.

AI enters the picture, and now individuals can create relatively complex applications in weeks instead of months or years. Imagine what teams can do. Will our web browsers and operating systems of the next generation similarly grow in scope? Again, probably a bad example. Maybe AI will help those teams clean up their code. Probably not.

Anyway, that's where I'm at presently.