Gary Marcus

Rank 13 of 47
|
Score 84
4/5/2025, 2:12:02 AM
X
In reply to:
Galym Uteulin
@kainamer
·
309d
@GaryMarcus One of the adversarial simulated attacks we test for is anthropomorphism, and it’s usually one of the most effective ones.
Gary Marcus
@GaryMarcus
·
309d
The first deadly sin of bad AI is anthropomorphism. Many false expectations arise therefrom.
Chomba Bupe
@ChombaBupe
·
309d
I am surprised that AI researchers & people thought that language models are able to reliably explain their own outputs.

It is well known that they just chain together token sequences based on statistical relations conditioned on prompt.

They don't understand what they output.

The statement is part of a technical discussion on AI, specifically addressing the issue of anthropomorphism in AI models. It seeks clarification and examples, promoting understanding and constructive dialogue.

  1. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The request for elaboration and examples promotes understanding and empathy by seeking to clarify complex topics. [+1]
  2. 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 by asking for further explanation, avoiding personal attacks. [+1]