optimalsolver

6 ms·

1. 211.▲
   by optimalsolver 2y ago

   >higher less qualified candidates Ironique.
2. 212.▲
   by optimalsolver 2y ago

   I recommend this article on the kinds of things people actually say on their deathbeds: https://www.theatlantic.com/family/archive/2019/01/how-do-pe... It usually doesn't make any kind of discernabl
3. 213.▲
   Mind Uploading And Continuity(selfawarepatterns.com)

   1 points by optimalsolver 2y ago|0 comments
4. 214.▲
   Can Knowledge Hurt You? The Danger Of Infohazards (And Exfohazards) [video](youtube.com)

   1 points by optimalsolver 2y ago|0 comments
5. 215.▲
   by optimalsolver 2y ago

   Also: https://xkcd.com/1425/
6. 216.▲
   by optimalsolver 2y ago

   Is there any actual evidence that the reasoning tokens output by current models actually represent the computation happening in the hidden layers? In both cases, the model is doing a ton of processing that you can't actually inspect, e
7. 217.▲
   by optimalsolver 2y ago

   There are plenty of self-taught people in the open source space making highly reliable software.
8. 218.▲
   ZebraLogic: On The Scaling Limits Of LLMs For Logical Reasoning(arxiv.org)

   1 points by optimalsolver 2y ago|1 comments
9. 219.▲
   The Big List Of Existing Things (2017)(everythingstudies.com)

   2 points by optimalsolver 2y ago|0 comments
10. 220.▲
    by optimalsolver 2y ago

    >This is a bitter lesson about the interplay between techlash activism and big tech power structures. Twenty years of privacy complaints have only made tech companies more powerful. So we should have done what, exactly, Ben?
11. 221.▲
    Compression, Prediction And Artificial Intelligence (2010)(fendrich.se)

    1 points by optimalsolver 2y ago|0 comments
12. 222.▲
    by optimalsolver 2y ago

    Hey, I'm interested in the details of this. How many persons in the puzzle? Did it include nested statements, conditionals and such? If the puzzle generator is hosted anywhere, I'd love to have a look at it.
13. 223.▲
    Compression Represents Intelligence Linearly (2024)(openreview.net)

    2 points by optimalsolver 2y ago|0 comments
14. 224.▲
    by optimalsolver 2y ago

    Thanks for the amusing mental image.
15. 225.▲
    by optimalsolver 2y ago

    Move 37.
16. 226.▲
    by optimalsolver 2y ago

    Big Tech Podcast listener?
17. 227.▲
    by optimalsolver 2y ago

    You're gonna need to ground those labels in something physical at some point. No one's going to let an LLM near anything important until then.
18. 228.▲
    by optimalsolver 2y ago

    >this $300Bn company Watch this space.
19. 229.▲
    by optimalsolver 2y ago

    > quick, intuitive decisions, focuses on what's novel/unknown Wouldn't it be the exact opposite? That is, novel stimuli requires more extensive processing at higher levels of cognition.
20. 230.▲
    by optimalsolver 2y ago

    But can o3 write a symphony? Seriously though, I'd like to hear suggestions on how to automatically evaluate an AI model's creativity, no humans in the loop.
21. 231.▲
    31 Laws of Fun (2009)(lesswrong.com)

    4 points by optimalsolver 2y ago|0 comments
22. 232.▲
    by optimalsolver 2y ago

    It's been found [0] that slightly varying Putnam problems causes a 30% drop in o1-Preview accuracy, but that hasn't put a dent in OAI's hype. There's absolutely no comeuppance for juicing benchmarks, especially ones no o
23. 233.▲
    by optimalsolver 2y ago

    >the public repeating the benchmarks themselves The public has no access to this benchmark. In fact, everyone thought it was all locked up in a vault at Epoch AI HQ, but looks like Sam Altman has a copy on his bedside table.
24. 234.▲
    by optimalsolver 2y ago

    >OpenAI has data access to much but not all of the dataset Their head mathematician says they have the full dataset, except a holdout set which they're currently developing (i.e. doesn't exist yet): https://www.reddi
25. 235.▲
    by optimalsolver 2y ago

    Uropian Eunion
26. 236.▲
    by optimalsolver 2y ago

    All episodes of the 90s Tintin animated series are on DailyMotion: https://www.dailymotion.com/video/x749uno
27. 237.▲
    by optimalsolver 2y ago

    Recursive self-improvement?
28. 238.▲
    by optimalsolver 2y ago

    >of all possible theorems almost none have short proofs Ok, but what about the theorems humans would actually care about?
29. 239.▲
    by optimalsolver 2y ago

    I guess I should expect it on a VC-tied forum, but it seems strange to talk about company valuations and investment outlooks in the event of an artificial superintelligence operating in the world.
30. 240.▲
    by optimalsolver 2y ago

    Considering how fast open source models are closing the performance gap, this is rather optimistic. Maybe if they somehow fixed hallucinations and kept the secret sauce to themselves, I could see them being worth that much, but all the top