Or has it just provided counterexamples to the way a problem is worded, and searched large but finite solution spaces?Or has it actually come up with new theories and structural explanations to obtain positive results?
>>17051067It solved protein folding.
>>17051071As I understand, it was just trained to predict the folding to decent but not perfect results, and doesn't explain whats physically going on.It just seems like AI is either nitpicking the wording of a conjecture or brute-forcing a solution. Not developing theories to explain why something fails (or even more importantly, when it doesn't fail) or how a solution ties in naturally with the theory.But i can't keep up with all the AI results so idk
>>17051071As i understand, it trained enough to predict the folding to a decent but imperfect degree, and offers no explanation as to whats going on or why.Like AI is just nitpicking and brute-forcing problems, no? Not tying things into the larger theory or figuring out not when things break, but why, and where they don't. Ykwim?
Woops replied 2x
Bot malfunction
>>17051071>It solved protein folding.AI psychosis caused you to hallucinate this. It makes guesses, a good deal of which end up nonphysical.
>>17051083It makes endless guesses and then verifys them. Thats basically intelligence my coping physicist.
>>17051095>It makes endless guesses and then verifys them.It doesn't. It can't. You're literally a raving psychotic patient.>Thats basically intelligenceIncoherent psychotic mumbling.
>>17051083It solved it well enough to win a Nobel prize. I'm sorry you feel threatened by rapid scientific progress. You wouldn't be the first, and we definitely need to find a way to help people like you cope with the pace of change before you break and start bombing data centres. Maybe we can get AI to solve that next.
>>17051118>It solved>itIs "it" in the room with us, psychotic patient?>solvedGuessing is not "solving".
Notice how the anti-AI fanatics can't actually articulate their positions in a coherent way, they just make bald assertions and call everyone else crazy.
>the anti-AI fanaticsTotally in the room with us right now, just like the "it" that won a Nobel Prize for solving protein folding.
>>17051067>moves the goal post yet again
>the Luddites (they're in the room with us right now, i swear) keep moving their official heckin' goalposts that they keep beaming into my head through the padded cell wallsEvery AI believer is mentally ill. No exceptions.
>>17051135If a human solved any major problem by guessing a counter-example, he would be called a genius, but when a machine does it suddenly it's not a "real" solution. AI could guess a cure for pancreatic cancer and the ideologues would be complaining that it didn't come with a new structural theory of pancreatic cancer at the same time.
>>17051141>If a human solved any major problem by guessing a counter-example, he would be called a geniusYes, because sheer autistic savant genius is the only way for a human to "guess" a counter-example to a major unsolved problem.>when a machine does itThe language model does not and cannot do "it" - whatever "it" is the hypothetical human would be doing - so your statement is a nonstarter.
>>17051141>AI could guess a cure for pancreatic cancerThen why hasn't it? That would be a positive outcome that's kinda hard to argue with, unlike what your handlers are doing with math.
>>17051149Because unlike mathematical results, medications need to go through expensive and strictly regulated clinical trials, and clinical trial approvals and funding are gatekept by committees of old farts like you who are suspicious of AI for illogical reasons and are holding back progress. We've already seen cases of individuals with the necessary financial and intellectual resources using AI to bootstrap highly effective boutique treatments for their own cancers. I'm betting that we'll start seeing cancer advances offered as a paid service in some legally friendly jurisdiction in the Caribbean long before any FDA approvals come.
>>17051205>because you can't bruteforce real lifeToo bad. In the meanwhile, it remains the case that "AI" is a net negative even from a purely pragmatic perspective, nevermind the fact that it's not intelligent.
>>17051210>I would rather die of cancer than admit that AI might be able to help me do something about itOK bro, you do you
>>17051215You realize your reply is incoherent, right? You're quite literally psychotic. You're replying to something a voice in your head said, not to my actual posts.
>>17051205you're right, instead of holding back progress they should run LLM trial-and-error iterations on groups of AI fans. identifying AI fan posts and tracing them back to the fan's real life identity would be a good use of the tech. you 100% should be on a government list, even if the medical trial idea falls through
>>17051071>>17051095>>17051118Google killed AlphaFold months ago, if you haven't noticed. They consider LLMs more important.
>>17051141>If a human solved any major problem by guessing a counter-example, he would be called a genius, but when a machine does it suddenly it's not a "real" solution.OpenAI had 10000 agents working 88 hours, that's literally 100 years. And all it could do was what people today were already in the process of working out. That's not impressive, that's not genius. And it didn't tell us anything new about Navier-Stokes we could follow up on.
