@scy I really hope it pops soon. But I don’t know anything anymore. Nothing makes much sense when it comes to the AI companies. And there’s still loads of government contracts that will be given to them.

@scy I'd say it's already pretty wobbly, and will continue to inflate and deflate for a while, until some time in '27 or '28 it'll suddenly pop. Like first slowly, then all at once.

@scy I do expect the bubble to pop, in that a bunch of people will lose money, companies will fail and it may lead to another horrible economic slump. One way or another the debt will get written off, hopefully before all those power guzzling data centres have been built. But then other actors will just take it up so I expect the problem of slop-generated everything will not go away.

@scy I hope.

@scy I think the financial side will pop loudly and with devastating effects.

Probably there will remain some LLM use in software development, mostly in a corporate setting, just like shitty frameworks have a lot of staying power there. But I don't see the mandatory use we see today persisting for long when the real costs fully manifest, although it will take time to propagate through the industry.

@scy I expect the AI bubble around the big players to pop, yes. The price for slop and low-effort throwaway jobs is too damn low.

LLM-generated code will stay, because of open weight models being run locally, I'm afraid.

@scy i hope 1, but fear 4.
I fear the biggest problem is that slop code is a (too) very slow ticking time bomb, which won’t go off until 1 to 2 years later, when AI can’t add something without breaking existing stuff and adding workarounds to fix that isn’t working anymore.