Post

i think a core beef that i have with a lot of this "automation is bad" sentiment lately is that there is no inherent moral good in laboring the story of humanity is one of invention, where we improve our conditions by building things that help us do more things more easily

tbh i find it deeply irritating and disingenuous to call this "automation", as if we'd just invented the dishwasher. it's not a contraption that performs a rote task; it's a box that spits out Whatever

the story of humanity is one of telling stories, and now i am constantly advertised the box as "What if you didn't have to tell stories. What if you could automate talking to your children. What if there were a TikTok nobody had to make. Just endless banal sludge, forever"

i think that this is a *little* bit of a waffles situation. i don't disagree that, for example, Sora is boring, bad sludge. but i wasn't talking about that in this thread i have a lot of beefs and concerns with llms! they just aren't what i was talking about here

you opened by characterizing the feelings of other programmers as "waffles are bad"

No, I was speaking about one specific dimension of LLMs. There’s a ton of shit here and I agree that a lot of it is bad. But in this thread I’m talking about one specific disagreement I have with a lot of people

Like: Elon should be in jail. This grok shit is abhorrent. I’m concerned about not being able to trust in reality any more Being concerned about psychosis is something I’m not sure is as widespread of an issue as some but is 100% of an issue to some degree

You could maybe make a locally running ~LLM that is exclusively usable for programming tasks and can't be used for any of the awful shit the general purpose models are used for but thats not the reality we live in right now.

I don't think there's a good way to ensure that local models can't be used for awful shit (although I think the WORST awful shit is the stuff coming out of the image and video models—don't get me wrong, you can do horrible things with text models, just. Less so than image/video), but...

i think there are gradients between these services, personally. i'd never give grok money. i give openai a little. i give anthropic a bunch. i can understand not wanting to give any of them any money