I don't think it's like the Industrial Revolution yet. I think it's hype, probably a bubble and we'll see another AI winter. We're seeing people vibe coding shitty games and holding it up as a kind of prophecy of what is soon to come... that we will be able to prompt the AI to build AAA games, which are codebases with millions of lines of C++, and many GBs of detailed art, sound, etc., all intricately put together by hundreds of people.

This is the best analogy of what is happening in software engineering as well. People are prompting the AI to code bouncing balls, or a landing page, and thinking it will soon be able to replace software engineers completely. I think there is no proof that this is going to be the case.

I really doubt that large codebases can be built without having an underlying capacity for logical reasoning and understanding... but to be clear, I'm not saying that AI will never get there... what I'm saying is that once it does get there, it's basically the singularity.

There is no intermediate step in which we let the AI code and somehow it won't be able to do literally every other job out there. We're in an interesting place. Either it's just a really good autocomplete and cool image generator, or we're all royally fucked, and we're about to hit the singularity, and we can say goodbye to the world as we know it.

I guess if you place a lot of faith in the ai assistants, it's easy to feel like this. My experience has been that every one I try is a waste of time, producing code that doesn't do what it's been promoted for, or worse, code with subtle bugs which take a keen eye and/or debugger to fix.

My experience with ai assistants has strengthened my identity as a maker/creator/builder because I see the slop that's produced in all the derivative works and want nothing to do with it.