Gary Marcus

Rank 19 of 47
|
Score 64
1/6/2025, 5:26:02 PM
X
In reply to:
Matthew Kolakowski
@MatthewKolakow9
·
189d
@GaryMarcus Yes, either it is a horn or a child with some form of sword like weapon mounted on his back. While I do not see a near term AGI path, I do believe that image generation will get to a point in the next 12 to 24 months where it stands up to scrutiny and matches realistic feature…
Gary Marcus
@GaryMarcus
·
189d
@MatthewKolakow9 real AGI presumably could do such thing, but i don’t see a near term path.

looks like the kid has a horn there, eh?
Matthew Kolakowski
@MatthewKolakow9
·
189d
@GaryMarcus I generated a similar image. I wonder how sanity and visual accuracy checks could be done post hoc? You could have a visual scan and comparison, but I wonder if that meets any kind of rigor or fact check criteria.
Gary Marcus
@GaryMarcus
·
189d
“Please generate a picture of a child riding a flying unicorn through a rainbow” - my kids, who note two tails, two horns, and one very weird looking rainbow.

[The point here is not that the correct output could not be generated, but rather that the system fails to sanity check…

The conversation discusses the challenges and future of AI image generation, focusing on visual accuracy and realism. It critiques current limitations and anticipates improvements.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The discussion is technical and does not harm individuals or groups. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The conversation promotes understanding of AI capabilities and limitations. [+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 dialogue about AI technology without personal attacks. [+1]