Considering how LLMs are trained, they prob contain a lot of sov cit stuff, wonder if a lawyer/judge can trick a LLM into going full sovcit by just adding a few words/rephrasing a bit.
Soyweiser
joined 2 years ago
No real understanding of what it's doing, it's just guessing.
Are they talking about the LLMs or the people who think just chatting with the LLM will fix it? :)
E: from a comment about this on hackernews:
Funniest PRs are the ones that "resolves" test failures by removing/commenting out the test cases, or change the assertions.
Perfect, no notes. Ship
Ben S voice "Look for the sake of the argument, what if we just assume im right, and ignore everything that shows I'm wrong. What then? Checkmate liberal!"
Yes, there I heard there is some javascript that various older versions of chrome/firefox don't properly execute for example. So you can use that to determine which version they are (as long as nobody shares that javascript with the public. So this might even not be javascript, I honestly know nothing about it just heard it).
That reminds me, remember there is an Xbox boycott going on for all the gamers out there. (Saw that after the boycot was started, both steam and humble pushed xbox game sales, the timing of which is very iffy).
Deleted earlier message, sorry I called Scott out for not doing things he had done. Even if the whole mods 'restricting her messages now only after she went after Scott' is quite iffy. (LW people write normally challenge failed "One upfront caveat. I am speaking about “Kat Woods” the public figure, not the person. If you read something here and think, “That’s not a true/nice statement about Kat Woods”, you should know that I would instead like you to think “That’s not a true/nice statement about the public persona Kat Woods, the real human with complex goals who I'm sure is actually really cool if I ever met her, appears to be cultivating.”" (The idea is good, this just reads like a bit of a sovcit style text and could have been replaced with 'I mean this not as an attack on her personally, I'm just doubting the effectiveness of her spammy posting style'). (E: I do agree with them however, not the 'we should check if this is effective' but more that the posting style is low effort, annoying, boring, dated, bit cringe etc).
Also: Scott: 'Mods mods mods, kat spill my jice help hel help help'
I was very tempted to go 'don't think it is more than one nobel guy, which is not great because of nobel disease anyway. I could link to rationalwiki here but that has come under threat because the people whos content you enjoy Scott started a lawsuit against them' but think that might be a bit culturewarry, and I also try not to react at the places we point towards. As that just leads to harassment like behaviour. Also Penrose is a Nobel prize winner who is against AGI stuff.
That is the one I was thinking of, the way the comments are phrased makes it seem like there are a lot of winners who are doomers. Guess Hinton is a one man brigade.
Yeah the financial illiteracy is quite high, on top of the rest. But dont worry AI nobel prize winners say it is possible!
(Are there multiple ai Nobel prize winners who are ai doomers?)
That gives me a 'you broke reddit' jackrobertsofficial is also empty for me (and empty if I use an incognito window, so I'm not blocked). I got the feeling that might be what was going on. Even if I had a hard time finding his old work, as the news articles he links on his own site were dead.
E: tried on my phone and it appears wtf, no wait. It is promoted, my addblockers just nuked it haha, my bad.
Look, im def on team Murderbot, but when ~~we~~ the AI's start building them I really hope Martha Wells gets some kickbacks at least.