Gary Marcus

Rank 19 of 47

Score 64

@EverydayAI_ For clarity: the point is not that you can get the correct output (that one isn’t even particularly hard), it’s that the system can’t sanity check its outputs, so errors, even absurd ones, are not uncommon.

(Also, I didn’t try hard, or even at all; it was in the first outputs my…

1/6/2025, 5:10:37 PM

In reply to:

Jordan Talks Everyday AI

@EverydayAI_

189d

If you try hard enough, Gary, you can get bad outputs from anything.

Kinda like if my goal was to travel from Chicago to LA and I decided to take a unicycle, that's probably a bad output.

(here's the same prompt one-shotting in Imagen.)

This post is engagement baiting and…

The statement critiques AI systems' inability to self-verify outputs, highlighting potential errors. It engages in a technical discussion about AI reliability, addressing public concerns about AI accuracy.

Principle 1:
I will strive to do no harm with my words and actions.
The statement aims to inform about AI limitations, aligning with the principle of doing no harm by raising awareness of potential issues. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
It promotes understanding by explaining a technical aspect of AI, fostering a more informed public. [+1]
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 engages in constructive dialogue, addressing a point made in the previous comment without personal attacks. [+1]