If AI writes the code, what do I stop learning?

A philosophical walk, downhill (if I fall, feel free to laugh). The question I’m carrying is simple: is this whole push towards automation, artificial intelligence and robotics heading in the right direction?

What the body learns

Let me draw a parallel with martial arts, though it works for anything: an instrument, a profession, cooking. When I study karate, I repeat a punch or a kick until it becomes mine. It ends up in muscle memory and in the nervous system, and from then on it just comes out. A cook does the same with a knife, a musician with an instrument.

We’re plastic: brain and body change shape with what we do. The simplest example is walking. A newborn can’t walk; then they learn, and stop thinking about it. So they think about jumping. Then they stop thinking about jumping and start playing with jumps. And so on.

Automation as imprinting

If the comparison holds, automating with robots and AI is an external version of what we already do inside ourselves: we take a process, make it automatic, and free up attention for the next level. It reminds me of the Italian saying “learn the craft and set it aside”. Set it aside where? In the body, in memory. Or, now, in a machine.

What if it’s wrong?

That’s exactly the point. If instead we were outsourcing things that end up making us dependent, inept or incapable, it would be the wrong bet: a road that ends at a cliff, and sooner or later we’d have to turn back.

Speaking for myself: I know a bit of Ruby, a bit of Java, a bit of C, each at its own level. If I stop writing code and let agents do it all, I stay at the high level: what the code should do, how, how robust and secure it must be. But I don’t touch the code anymore. Six months from now, if I get my hands back in it, I’ll probably be close to useless. I touched on this when talking about the evolution of the developer in the AI era.

One muscle weakens, another grows

My counterargument is: fine, but I’ll always have this tool for writing code. That muscle may weaken, but another one gets trained: the big picture, problem solving, managing problems and explaining to other people how to solve them. And maybe that’s more useful across the board.

I don’t really have the answer. It’s an example of something that can be negative and positive at the same time.

What are you automating? What good and bad is AI bringing to your field? Let me know in the comments on the video.