This is fun to think about. I used to think that all software was largely garbage, and at one point, I think this _was_ true. Sometime over the last 20 years, I believe this ceased to be the case. Most software these days actually works. Importantly, most software is actually stable enough that I can make it half an hour without panic saving.
Could most software be more awesome? Yes. Objectively, yes. Is most software garbage? Perhaps by raw volume of software titles, but are most popular applications I’ve actually used garbage? Nope. Do I loathe the whole subscription thing? Yes. Absolutely. Yet, I also get it. People expect software to get updated, and updates have costs.
So, the pertinent question here is, will AI systems be worse than humans? For now, yeah. Forever? Nope. The rate of improvement is crazy. Two years ago, LLMs I ran locally couldn’t do much of anything. Now? Generally acceptable junior dev stuff comes out of models I run on my Mac Studio. I have to fiddle with the prompts a bit, and it’s probably faster to just take a walk and think it over than spend an hour trying different prompts… but I’m a nerd and I like fiddling.
I ended up watching this talk this evening:
> Jonathan Blow - Preventing the Collapse of Civilization (English only) :: https://inv.nadeko.net/watch?v=pW-SOdj4Kkk
and it made me think of your comment. In summary, I disagree, and think that video argues the point very convincingly.