Gary Marcus

Rank 20 of 47
|
Score 39
Gary Marcus
@GaryMarcus
·
574d
So many people are confused about the relation between human cognitive errors and LLM hallucinations that I wrote this short explainer:

Humans say things that aren't true for many different reasons
• Sometimes they lie
• Sometimes they misremember things
• Sometimes they fail
11/4/2025, 11:59:20 PM
X
In reply to:
Elan Barenholtz, Ph.D.
@ebarenholtz
·
11d
@GaryMarcus Humans also hallucinate. We misremember, confabulate, and even generate false memories. The difference is a quantitative one, not a qualitative one and it can likely be attributed to the very different training regimens that humans and LLMs undergo. In particular, humans undergo
Gary Marcus
@GaryMarcus
·
11d
@ebarenholtz i think this is backwards. LLMs hallucinate far more than people because they lack storage and retrieval.
Elan Barenholtz, Ph.D.
@ebarenholtz
·
12d
LLMs suggest that memory and knowledge aren’t facts stored in the brain. They are generative capacity. To “remember” where you left your keys is to be able to say so, reach for them, visualize them there. These capacities don’t depend on knowledge; they ARE the knowledge.

The statement is part of a public discourse on the comparison between human cognitive errors and LLM (Large Language Model) hallucinations. The conversation involves a substantive engagement with public issues related to artificial intelligence, cognitive science, and the nature of memory and knowledge.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm with words or actions, as it is part of an academic discussion. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement contributes to understanding by engaging in a dialogue about the differences between human and AI memory processes. [+2]
  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.
    The statement engages in constructive criticism and dialogue, responding to differing opinions without personal attacks. [+2]
  4. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement upholds free speech principles by responsibly engaging in a public debate on AI and cognitive science. [+1]