Post #3000931

2026-01-29 00:34 UTC

@krig@goto.liten.app @sarna@types.pl > The article keeps saying "assuming the utility is there" - well, is it? Assuming that's true let's me make a stronger point about when to use AI. I'm saying even if AI does what it's supposed to and has no negative externalities or societal harms, *even then* there are still many places where I wouldn't want to use it. If we often shouldn't use the best possible AI we can imagine, we should certainly use the worse AIs we actually have even less.

Replies (1)

@krig@goto.liten.app 2026-01-29 06:10

@munificent@mastodon.social @sarna@types.pl Oh I completely agree with you; personally even if it does what it's supposed to I would not touch it. I genuinely think it is inducing brain damage in a lot of developers right now. People don't seem to realise that the way you grow and learn something is by doing it, and if you outsource the doing not only will you stop learning but your abilities will absolutely atrophy. It's as if all the other joggers out there are suddenly riding little golf carts around the park, bragging about how much faster their jogs are. BUT once again - assuming it's true, assuming there is utility. I feel like even that is giving it too much credence! I just don't see the claimed utility. The only example of someone who uses it and is demonstrably productive that I have seen is Mitchell Hashimoto.. and despite him saying that he benefits I really don't think he is more productive with it than without it. Somehow he built his career before LLMs doing exactly the same kind of things as he is doing now with the same level of output.

Open ##3000932