Gary Marcus

Rank 13 of 47
|
Score 84
6/27/2025, 4:47:24 PM
X
In reply to:
Geoffrey Miller
@primalpoly
·
228d
@GaryMarcus Haha, some current AI systems make silly errors.

But how exactly is that any counter-argument against the strong takeoff theory that Artificial General Intelligence -- if and when it's ever developed -- could rapidly self-catalyze into Artificial Superintelligence?
Gary Marcus
@GaryMarcus
·
228d
The most amazing thing about the “strong takeoff theory” of superintelligence that had everyone in a panic was that it turned out to be true.

Exactly two weeks after Imagen 4 drew the diagram below of a dog, Imagen 4 took over the Earth.

A week later, and it had colonized the
yoda360
@vperni360
·
229d
@GaryMarcus Imagin-4 lol. Looks like VLMs can't label their own creation... Architecture issue obviously.. pixel grid n lost track 😂

The statement engages in a discussion about the feasibility and timeline of AI achieving superintelligence, specifically addressing the 'strong takeoff' theory. It presents a skeptical view on the immediacy of such an event, suggesting that even with rapid advancements, a superintelligent agent would require time for empirical research and resource gathering.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm and contributes to a reasoned debate about AI development. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    It promotes understanding by providing a counterpoint to the idea of an imminent strong takeoff. [+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.
    Engages in constructive criticism by challenging the immediacy of the strong takeoff theory without personal attacks. [+1]