dwarkesh_patel
Region: US
Thursday 27 November 2025 21:02:23 GMT
14123
610
14
31
Music
Download
Comments
Mika :
These short videos are way better for non engineers like me. I tried the whole video but its too technical for me
2025-11-28 12:45:39
18
__kyledrogo__ :
Isn’t RLHF a version of the post training learning he’s describing? Not pure RL, but fits the bill
2025-12-02 17:10:34
0
Marty :
The biggest thing i disagree with this is the determination to mimicing human intelligence. Ai can at this point evolve into something even more intelligent, and possibly even a new intelligence. But what is intelligence. I coin a new intelligence. Digital intelligence
2025-11-28 19:40:46
0
bikostevie :
Pattern matching and retrieval from a large source of data at scale is not AGI and definitely not close to human intelligence. We are struggling with mere image generation, so what makes us think we can capture ALL HUMAN SENSES needed to interact with the physical world?
2025-11-29 03:06:59
0
Lasermike026 :
He’s on to it!
2025-11-28 05:35:13
1
Jon :
The level of articulation! Top.
2025-11-29 19:29:44
2
threebod :
huh? it’s the opposite. “narrow” ai was an attempt to explain why we had super intelligence within a field like chess. general came first.
2025-11-28 21:27:50
1
ai.is.nuts :
IIIC the optimal superintelligence is a labyrinth of narrow intelligences — each achieved, not via drop-in supported by pre-training (AGI style), but via continual learning.
2025-11-28 05:30:07
0
Abity :
the only reason the term AGI exists, I think, is because without a huge ass promise of nothing, no huge capital raising 😂😂😂😂😂😂
2025-11-28 16:39:18
0
Odelbar :
Spend many minutes saying absolutely nothing....
2025-11-28 13:51:45
1
To see more videos from user @dwarkesh_sp, please go to the Tikwm
homepage.
