sc_griffith @ sc_griffith @awful.systems Posts 15Comments 394Joined 2 yr. ago
at least make a pretense of having read the article instead of very obviously reacting to the headline jfc
one of the best articles I've ever read was an inch by inch teardown of the entire concept of mass produced lab grown meat: https://thecounter.org/lab-grown-cultivated-meat-cost-at-scale/ . it's never, ever going to work at scale and I'd go so far as to say it's the food equivalent of all the usual tech grifts we talk about here
they don't think it's a messy and inevitable fact of the world that certain people have outsize wealth. they think it fucking rules
hate this feeling and yet that's also a big part of why I keep tabs on them. they're there whether we choose to look at them or not
this is... your fault...
just found out there is genius.com annotation for this
https://genius.com/Reddit-copypastas-uwu-whats-this-annotated
I can't tell you how happy it would have made me to feel abandoned as a kid
looking up uwu on know your meme "uwu," what is this
I see. so "we only sent the registrar a report about one page" is total bullshit
will a domain registrar actually take down one page at a time instead of an entire domain? genuine question idk how Internet works
this guy nailing the UHC CEO is the only good thing they will ever have a chance of claiming credit for, so of course they rush to disavow him
reddit just launched an LLM integrated into the site. hardly any point going through what garbage these things are at this point but of course it failed the first test I gave it
for millions of years humans lived in caves. in the last year alone we've mastered automated child grooming, learned to boil lakes to generate glue pizza recipes, and for the first time created fully artificial asshole reply guys. progress is exponential
now I'm thinking about how ben carson is considered to be an almost supernaturally skilled surgeon and yet is ben carson
gave it a quick skim. I lack any relevant background. the bit they push most seems to be that you can improve the performance of error detection tools by determining the most important tokens in an answer and running your tools on the tokens near those. this seems to be in contrast to absurdly naive approaches like averaging the tokens (???) or just looking at the last token of the response (????).
what are the most important tokens? they're the ones that change the factuality of the answer if you change them. how do you determine that? you don't, lmao. you just ask an LLM what the most important words are
what are the error detection tools? you will never guess
mirror universe ed zitron like
for a post from an AI company this is surprisingly honest about the general helplessness in defeating so called hallucinations
https://www.kapa.ai/blog/ai-hallucination
probably a good sign for even boosters to be talking this way
that's gonna be his new thing. 30 blog posts in a row dunking on godot
really into this whole far right coup and far right right hegemony thing a lot of countries are trying out rn. really cool
I have a stanford degree just like this guy btw. so you have to take my idea seriously