Gary Marcus

Rank 11 of 47

Score 83

@1carringtone @lmarena_ai @liron i think we can safely conclude what I said: you can never be sure.

i didn’t say all models make that error all the time. (though it’s second such report i have seen in a few days).

is there any model that doesn’t sometimes make boneheaded errors like that (which I like to call

5/24/2025, 6:06:13 PM

In reply to:

Kinyanjui Carrington 🇰🇪🇬🇭🇿🇦 ⏯️

@1carringtone

253d

@GaryMarcus I know Twitter isn't the best place for "science", but concluding this requires taking LLMs on the @lmarena_ai leaderboard and asking them this question.

Otherwise we're just judging complex systems by anecdotal failure modes

Loved your @liron interview. You had great points

Gary Marcus

@GaryMarcus

253d

2.5 years later, you can still never be sure anything an LLM tells you is true.

The statement discusses the reliability of language models, acknowledging their potential for errors. It engages in a constructive dialogue about the limitations of AI, referencing specific examples to support the argument.

Principle 1:
I will strive to do no harm with my words and actions.
The statement does not harm but rather highlights the importance of understanding AI limitations. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
It promotes understanding by discussing the nature of AI errors. [+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 responding to previous comments and providing examples. [+1]