Gary Marcus

Rank 14 of 47
|
Score 83
11/23/2025, 5:21:35 PM
X
In reply to:
David Manheim
@davidmanheim
·
9d
@GaryMarcus @primalpoly @eli_lifland I'm less pessimistic than @primalpoly, but am confused about why people see current approaches as more dangerous than others that are often proposed. (At least with LLMs, we can hope they'd share some basic human ethical drives by default- which seems untrue of other approaches!)
Gary Marcus
@GaryMarcus
·
10d
@primalpoly @eli_lifland if we can’t build in stable constraints i agree we are in trouble.

and yes let’s talk soon.
Geoffrey Miller
@primalpoly
·
10d
@GaryMarcus @eli_lifland Someday, Gary, I aim to convince you that there is no possible path towards AGI/ASI that would be 'safe' in the sense of 'plays nice with humanity for many generations'.

In fact, it's utterly baffling to me that any serious person could think that any agentic AGI/ASIs capable of
Gary Marcus
@GaryMarcus
·
10d
In many ways aside from the word “soon”, this AI timeline analysis from @eli_lifland is actually very similar to my own take.

• Absolutely agree that AGI and ASI will eventually be built and will be wildly transformative

• Absolutely agree that we are not ready for it.

My
Eli Lifland
@eli_lifland
·
11d
We’re highly confident that:
a) AGI and ASI will eventually be built and might be built soon.
b) ASI will be wildly transformative.
c) We’re not ready for AGI and should be devoting much more attention to preparing for it.

We’re **not** highly confident that:
1) AGI will arrive

The statement is part of a broader discussion on the safety and ethical implications of developing AGI (Artificial General Intelligence) and ASI (Artificial Superintelligence). The tone is assertive and somewhat dismissive, indicating a strong belief that current approaches to AI safety are insufficient.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly cause harm, but its dismissive tone could discourage open dialogue, which is important for collaborative problem-solving. [-1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement lacks an effort to promote understanding, empathy, or compassion, as it dismisses the possibility of achieving safety with current methods without offering constructive alternatives. [-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 statement does not engage in personal attacks, but it does not foster constructive criticism or dialogue, as it ends the conversation with 'end of story,' which can shut down further discussion. [-1]