The fact is, LLMs are incapable of invention and synthesizing new ideas. They can't contribute to the zig compiler because they have not been trained on the zig compiler, because it doesn't exist yet.
Yes, they can churn out simple apps, and quickly. That's a pretty useful thing, especially for people that don't know how to write code. But that's not as revolutionary as you think it is.
Others have mentioned the hype around 3D printing several years back. Kinda the same story. People thought manufacturing was dead, stores were dead. You'd just print everything you need yourself! Turns out it's not quite like that. It could still get to that point someday but these are hard problems that take time.
Similar with LLMs. It took us, what, 70 years of computer and AI research to get to this point? And people assume we're going to skyrocket way past this point in another year or two?
