Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

Gary Marcus Marcus on AI Published

“Scale Is All You Need” is dead

We recorded 9 claims from this post. 3 were checked against evidence. 6 can’t be checked. 9 more are held. Read the post at its source.

Claims from this post (9)

In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.

  1. Assessment: SupportedSaid

    “And now, at last, the field is starting to realize it.”

    What we tested: The AI field is now starting to realize that scaling alone will not reach AGI.

    What happened: A survey of 475 AAAI community members found that 76% thought scaling current systems was unlikely to produce AGI. (The Decoder, )

  2. Assessment: SupportedSaid

    “The machine learning community, which has unwelcomed me since I started vocally critiquing the scaling approach, has finally started to come around.”

    What we tested: The machine learning community has started to accept critiques of the scaling approach.

    What happened: An industry write-up of NeurIPS 2025 says the community is moving away from 'just scale it'. Published a week after the post, but it describes the conference held before it. (Lambda (company blog), )

  3. Assessment: Can’t be checkedSaid

    “It all works, until it doesn’t. We desperately need more reliable technology.”

    What we tested: LLM-based tools are unreliable, and more reliable AI technology is badly needed.

  4. Assessment: Can’t be checkedSaid

    “[Scaling] may well be better than anything else we currently have, but the fact that it still doesn’t really work, even after all the immense investments that have been made in it, should give us pause.”

    What we tested: Scaling may well be better than anything else we currently have, but because it still doesn't really work after all the immense investment in it, that should give us pause.

  5. Assessment: Can’t be checkedSaid

    “Knowing what to borrow and what not is likely to be more than half the battle.”

    What we tested: In building AI, knowing what to borrow from human and animal cognition, and what not to borrow, is likely to be more than half the battle.

  6. Assessment: Can’t be checkedSaid

    “But it should look to human (and animal cognition) for clues.”

    What we tested: AI research should look to human and animal cognition for clues, without slavishly replicating it.

  7. Assessment: UnresolvedSaid

    “If we are to build AGI, we are going to need to learn something from humans, how they reason and understand the physical world, and how they represent and acquire language and complex concepts.”

    What we tested: If we are to build AGI, we are going to need to learn something from humans: how they reason and understand the physical world, and how they represent and acquire language and complex concepts.

  8. Assessment: Can’t be checkedSaid

    “Let’s bring in the cognitive scientists, and stop fantasizing that data and compute will solve all our problems.”

    What we tested: The AI field should involve cognitive scientists and stop believing data and compute alone will solve its problems.

  9. Assessment: Can’t be checkedSaid

    “The time for neurosymbolic AI and world models and causality is now.”

    What we tested: AI research should now pursue neurosymbolic AI, world models and causal reasoning.

Held from this post (9)

These records did not pass every check, so nothing about them is published. Held is not a finding.