Gary Marcus

Rank 19 of 47
|
Score 64
6/16/2025, 5:33:44 PM
X
In reply to:
Gary Marcus
@GaryMarcus
·
72d
@IAmTimNguyen no, i am just clarifying what i said.

there is no current solution.

anything else is speculative

since i pointed out related problems in 2001, and have seen no principled solution, i am not optimistic about having robust solutions in 2026. eventually one may come, but i think
Timothy Nguyen
@IAmTimNguyen
·
72d
@GaryMarcus "No solution to this problem" is quite unfair as it is the opposite to what Terry says. "we will see where things go" is quite the about-face Gary!
Gary Marcus
@GaryMarcus
·
72d
@IAmTimNguyen it is a report of where AI is now, and it is fair.

we will see where things go.
Timothy Nguyen
@IAmTimNguyen
·
72d
@GaryMarcus This is unfair. Terry Tao has suggestions and a vision for how AIs can be used successfully, which he even eludes to this in the very same video clip. Elsewhere, Terry writes:

"I expect, say, 2026-level AI, when used properly, will be a trustworthy co-author in mathematical
Gary Marcus
@GaryMarcus
·
74d
Terence Tao reporting on the same problem we have seen endlessly in other domains: LLM’s produce output that “looks” correct but is often deeply wrong and even stupid on careful inspection.

I know of no domain in which this is NOT the case.

And yet people seem surprised over
vitrupo
@vitrupo
·
74d
Terence Tao says today's AIs pass the eye test -- but fail miserably on the smell test.

They generate proofs that look flawless. But the mistakes are subtle, and strangely inhuman.

“There's a metaphorical mathematical smell.. it's not clear how to get AI to duplicate that.”

The conversation involves a discussion on the current state and future of AI, particularly in relation to its ability to produce reliable outputs. The participants are debating the potential for AI to achieve AGI (Artificial General Intelligence) and the challenges it faces, such as producing outputs that appear correct but are fundamentally flawed. The tone is analytical and speculative, with differing opinions on the progress and future of AI.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The discussion is largely respectful and focused on the technical challenges of AI, aiming to do no harm. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The conversation promotes understanding by discussing the complexities and limitations of AI, though it could benefit from more empathy towards differing viewpoints. [+1]
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The participants engage in constructive criticism, though there are moments of disagreement that could be more collaborative.