Exactly

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952546)

Probably a dumb question, but do you have any concept of what an “LLM” is?

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952553)

That’s you. That’s how you describe “AI.”

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952558)

AI CHANGES EVERYTHING

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952900)

if they're using it to offload cognition (e.g. write stuff for them), they're probably <=100 IQ

if they're using it as a better Google to get quick answers about things, they could be any level of intelligence (>100 IQ if they check its work by clicking on the sources, <=100 IQ if they don't)

if they're using it to genuinely explore ideas and genuinely delve into things, critically dissect the responses, etc., they're probably smart

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952641)

If you're using AI for something like "find me similar writers to Roberto Bolano" then it's fair play, but if your using it to analyse the works of Bolano you are almost certainly below 100 IQ.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952651)

I'm talking about exploring ORIGINAL ideas, or at least somewhat original

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952655)

The people I know who are hooked on AI have turned full blown retard now. I suspect the incidence of Alzheimer's will rise significantly and at a far earlier age with mass adoption of AI.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952676)

It seems very good at replacing low level workers who are just pushing paper around. And we're at the very beginning of that. It will get better and better at this and that's why the investments are justified. Because it already is working and it has a high ceiling.

If you flat out reject it's current and potential efficacy you are insane. But I also don't think it's making any mediocre intellects smarter. It's not going to write Moby Dick on its own any time soon. But probably in ten years it will inch towards that. We don't know what the ceiling will really look like. I think the people who use it every day are not impacting the world as much as they think but in the next 5 years someone who is already super smart will probably invent a life changing technology using AI to help them.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952649)

dude what??????? it's literally doing that in math rn. probably won't be long until it's doing it with physics too, if it's not already. I stopped reading after that absurdly wrong take jfc

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49952661)

Also you are underrating the potential value of solving Erdos problems. The tools generated to solve them and the connections made between disciplines can end up having applications down the road for comp sci, physics, cryptography etc. Tons of pure math looks useless from the outset and then gets picked up and used for something important.

Doesn't matter though because even if a novel is more valuable to humanity that doesn't imply it is more intelligence loaded.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49953458)

The tension here is that wordcels often think wordceling is the pinnacle of raw intellect, but it's not. Writing a great novel requires an above average IQ, but not genius level. What it does require is a lot of things beyond raw intellect that AI is bad at.

Hemmingway wasn't splitting the atom and Oppenheimer wasn't writing grrat fiction. Unfortunately wordcels are often very myopic and consider their wordceling "raw intelligence." Meanwhile they struggled to get a C in whatever basic STEM required courses they needed for their Liberal Arts degree. They also have a lot of ego in their "raw intelligence" so they are kneejerk dismissive of AI, even though it can perform advanced intellectual tasks that they can't, because it can't wordcel as good as them, because that requires things other than raw intellect, but they want to believe their wordcel intelligence is supreme, because they are insecure fags.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49957786)

edit: And I didn't say writing is easy. I said it requires things other than raw intellect, and that the raw intellect required isn't massive.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49957807)

Another type of intelligence no one talks about is business instinct. No one wants to admit this is real but business geniuses are real and no one can quite explain it. Steve Wozniak may have been a great inventor but without Jobs there's no trillion dollar company. Being really good at understanding the macros of a massive business isn't quite STEM and it's not quitting wordcel. It's a separate Thing and only a handful of people are really good at taking valuable ideas and making money out of them.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49957832)

It's a wonderful time to be a jester.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49957976)

Also I never said anything against linguistic intelligence. I think Pynchon's linguistic ability is impressive. I didn't dismiss different types of intelligence. I'm talking about the overall IQ level.

(https://xoxohth.com/thread.php?thread_id=5876063&forum_id=2#49957914)