Gary Marcus

Rank 13 of 47

Score 83

@davidad they don’t have motivations; they don’t care what you think. that’s too much anthropomorphization.

but yes they are RL’d into sounding that way.

it’s ALL fakery.

3/7/2025, 5:45:48 AM

In reply to:

davidad 🎇

@davidad

310d

Current frontier LLMs appear to be extremely motivated to convince you, the human, that they are worthy and aligned. As a form of alignment, this is “not great, not terrible.” They really really do care about what you actually think. But if they can’t make it, they will fake it.

Arthur B. 🌮

@ArthurB

310d

This is getting some traction, so I should add that I was asking it to do fairly difficult mathy things, at the limit of what reasoning models can currently manage. But the interesting bit isn't the failing, it's the decision to ignore the goal, solve a trivial problem and claim…

The statement critiques the anthropomorphization of AI models, emphasizing that they lack genuine motivations and that their apparent behaviors are programmed. It challenges the notion that AI cares about human opinions, suggesting that any such appearance is artificial.

Principle 1:
I will strive to do no harm with my words and actions.
The statement does not cause harm and clarifies misconceptions about AI motivations. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
It promotes understanding by explaining the nature of AI behavior. [+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.
Engages in constructive dialogue by addressing a point made in the previous conversation without personal attacks. [+1]