Gary Marcus

Rank 19 of 47
|
Score 64
6/15/2025, 3:18:23 AM
X
In reply to:
Gh🫥💲T🕳️🅿️i🦾🅾️🎚🕳️🎹D
@Ghost_Pilot_MD
·
21d
@GaryMarcus @Montreal_AI @ceobillionaire Thank you. Will check it out
https://www.reddit.com/r/Machine…
Gary Marcus
@GaryMarcus
·
21d
@Ghost_Pilot_MD @Montreal_AI @ceobillionaire Haven’t seen the multi-agent paper; orchestration is important. Doing it right requires symbols.

I think neurosymbolic is necessary but not sufficient, see eg Next Decade in AI for discussion
Gh🫥💲T🕳️🅿️i🦾🅾️🎚🕳️🎹D
@Ghost_Pilot_MD
·
21d
@GaryMarcus Dr. Marcus…. genuinely curious.

In your recent o3 paper, you highlight how LLMs miss core hierarchies like phrase structure and binding. Do you see symbolic or neurosymbolic models as the ONLY viable path going forward to fix these issues?

What @Montreal_AI (@ceobillionaire)
Gary Marcus
@GaryMarcus
·
21d
If we were 2 years from some kind of singularity-like AGI, I would expect current AI to be able to

• (minimally) do essentially anything cognitive that a bright 10 year child could do, such as understand movies, acquire the basics of new skills quickly, learn complex,

The statement is a lighthearted acknowledgment of a Reddit thread related to a discussion on AI. It does not engage substantively with the public issue being discussed, which is the future of AI and its development.