\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

6/9/25 AI thread

https://x.com/EHuanglu/status/1931395020084666757 there's...
Aromatic philosopher-king
  06/09/25
a good use is turning Youtube and other video data into 3d e...
zippy fantasy-prone dopamine
  06/09/25
i've read about this but i don't think it will actually work
Aromatic philosopher-king
  06/09/25
...
Violent pink milk
  06/09/25
https://garymarcus.substack.com/p/a-knockout-blow-for-llms ...
Aromatic philosopher-king
  06/09/25
...
Violent pink milk
  06/09/25
...
Boyish Fluffy Depressive Roommate
  06/09/25
Surprise, surprise, the “A.I. revolution” is a b...
puce underhanded therapy
  06/09/25
transformers saturated most benchmarks at human level or abo...
zippy fantasy-prone dopamine
  06/09/25
I trained a local LLM on xo and it called me a fag
Transparent Incel
  06/09/25
...
Heady domesticated new version
  06/09/25
https://x.com/levie/status/1931914258528477687 Doctors we...
Aromatic philosopher-king
  06/09/25


Poast new message in this thread



Reply Favorite

Date: June 9th, 2025 9:43 AM
Author: Aromatic philosopher-king

https://x.com/EHuanglu/status/1931395020084666757

there's literally no commercial application for this except porn lmao

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#48999060)



Reply Favorite

Date: June 9th, 2025 12:32 PM
Author: zippy fantasy-prone dopamine

a good use is turning Youtube and other video data into 3d environments for training robots that can then be rapidly adapted to real world environments.

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#48999490)



Reply Favorite

Date: June 9th, 2025 3:12 PM
Author: Aromatic philosopher-king

i've read about this but i don't think it will actually work

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000084)



Reply Favorite

Date: June 9th, 2025 3:13 PM
Author: Violent pink milk



(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000089)



Reply Favorite

Date: June 9th, 2025 12:21 PM
Author: Aromatic philosopher-king

https://garymarcus.substack.com/p/a-knockout-blow-for-llms

wow

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#48999444)



Reply Favorite

Date: June 9th, 2025 3:14 PM
Author: Violent pink milk



(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000091)



Reply Favorite

Date: June 9th, 2025 3:24 PM
Author: Boyish Fluffy Depressive Roommate



(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000113)



Reply Favorite

Date: June 9th, 2025 4:00 PM
Author: puce underhanded therapy

Surprise, surprise, the “A.I. revolution” is a bunch of snake oil marketing pushed by a private equity industry desperately in need of the “next big thing” after the Fed spiked interest rates in ‘22. News at 11.

I would encourage anyone concerned about “AGI” destroying the world or shitty chatbots “ending employment” to please watch the PBS Frontline special Age of Easy Money.

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000198)



Reply Favorite

Date: June 9th, 2025 5:20 PM
Author: zippy fantasy-prone dopamine

transformers saturated most benchmarks at human level or above in a few years and are still rapidly improving. a paper comes out showing limitations of this one architecture and you conclude that the machine learning craze is a scam - what?

the fact that current models are already as capable as they are and have weird failure modes like this makes me more concerned, not less. there are tons of ideas out there still remaining to be tested. transformer based LLM with chain of thought training is hardly the only game in town. this doesn't provide much evidence for the idea that a radically new paradigm is necessary.

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000380)



Reply Favorite

Date: June 9th, 2025 12:22 PM
Author: Transparent Incel

I trained a local LLM on xo and it called me a fag



(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#48999447)



Reply Favorite

Date: June 9th, 2025 12:29 PM
Author: Heady domesticated new version



(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#48999471)



Reply Favorite

Date: June 9th, 2025 3:28 PM
Author: Aromatic philosopher-king

https://x.com/levie/status/1931914258528477687

Doctors went from 75% accuracy without AI to 85% accuracy when using AI for diagnosis. In the future it will be malpractice for your doctor to not be using AI. This is such a clear societal benefit, that it alone should cause everyone to be much more pro AI progress.

(http://www.autoadmit.com/thread.php?thread_id=5735202&forum_id=2E#49000125)