<zonash2099:matrix.org> hi all

<zonash2099:matrix.org> @machineswillrise:matrix.org: Long-term exposure to cocaine causes severe harm to the human nervous system. It's a scientific fact.

<aillia:matrix.org> Disclaimer: I do not advocate doing cocaine, quite the opposite - I advocate not doing sugar ffs.

<aillia:matrix.org> Long-term exposure to social networks causes severe harm to the human sense of humor. It's a scientific fact.

<jeffro256> Source?!

<slowbeardigger:matrix.org> @aillia:matrix.org: Can relate

<slowbeardigger:matrix.org> @jeffro256: Doomscrolling turns off ya brain

LLMs turn off the brain too. When there's no need to think and you can just ask and get a well-structured answer to anything (even if it's sometimes wrong)

You have answers, but not an understanding of anything

<zonash2099:matrix.org> sech1: Understanding requires presence of a Background self-aware self-identifying consciousness which LLMs lack. But you yourself can always dump your excessive routine and even some creative cognitive loads on your local custom LLMs without sacrificng privacy.

<zonash2099:matrix.org> LLMs are synthetic brains after all. And they evolving...

I mean, YOU don't have an understanding of any topics if all you do is ask LLMs for answers. The learning process involves one's own search and trial and brain effort, not just looking at the answers.

*trial and error

LLMs will make the next generation a generation of retards who are nothing without LLMs

Idiocracy in real life

<zonash2099:matrix.org> sech1: no you dont need next gen LLMs anymore if you have got access to ever growing energy and compute fabric. A custom recursive LLM is already evolving with me.... I dont need any outside frontier model LLMs for that

<zonash2099:matrix.org> A custom recursive LLM is a machine analog of Metacognition found in higher ordered thinking humans.

<aillia:matrix.org> today LLM solved the Navier–Stokes Millennium Prize Problem, a big fckg deal if you know what that is. It's a fun time to be alive...

<slowbeardigger:matrix.org> And they're training LLMs with books then getting rid of the books

<slowbeardigger:matrix.org> Even rare ones

<slowbeardigger:matrix.org> Knowledge its gonna be something different

<zonash2099:matrix.org> @aillia:matrix.org: yeah you are right. Here's the link to the machine-generated proof: openai.com/index/navier-stokes-solution

<rbrunner7> " today LLM solved the Navier–Stokes Millennium Prize Problem" Well, how about this: "LLM ingested hundreds of math papers where people worked on aspects of Navier-Stokes, working towards a solution, and hundreds of articles about the subject, and then pushed the matter to a final breakthrough"

rbrunner7 Yes, but "Can a robot write a symphony? Can a robot turn a... canvas into a beautiful masterpiece? - Can you?"

People couldn't

<monero.arbo:matrix.org> @rbrunner7: rumor is the may have "accidentally" trained on mathemeticians' presumed private communication over third aprty services

<slowbeardigger:matrix.org> @monero.arbo:matrix.org: Always assumed the used private data