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

“The history missed is the the fact that many neural net enthusiasts, have argued against symbols for decades, and the reality that is missed is the above-documented fact that symbolic descriptions like “red cube on blue cube” still elude the state-of-the-art in 2022.)”

The New Science of Alt Intelligence, paragraph 48. The full statement is at the source.

Our reading, the sentence that was checked:

Many neural net enthusiasts have argued against symbols for decades, and symbolic descriptions like 'red cube on blue cube' still elude the state of the art in 2022.

Assessment: Supported Evidence as of

Where and when Gary Marcus wrote it

Marcus on AI, The New Science of Alt Intelligence, paragraph 48.

Said
Source observed
live page, full text, public
Evidence as of
the cutoff for this assessment
Assessment recorded
assessment marcus-the-new-science-of-alt-intelligence-fdddcc-a2

We recorded 20 claims from this post. See every claim from this post.

The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2022-05-14. This page does not claim the wording was captured then.

What the claim commits to

Kind
Observation
Deadline, as the speaker wrote it
“in 2022”
Deadline state when assessed
No horizon
Deadline state on 2026-10-29
No horizon (the release’s reference date)

From the speaker

  • many neural net enthusiasts
  • argued against symbols for decades
  • symbolic descriptions like “red cube on blue cube”
  • still elude the state-of-the-art in 2022

Our interpretation

  • The historical clause concerns a substantial strand of connectionist opposition to symbolic accounts of cognition; it does not claim unanimity among neural-network researchers.
  • Decades means that this position is documented over a period spanning at least twenty years, without requiring the same individuals to maintain it throughout.
  • In the surrounding discussion of unreliability, elude means that leading models could not reliably satisfy such compositional descriptions, not that no successful image could ever be generated.
  • The target period includes the unspecified earlier historical record through publication. The contemporary-performance clause is restricted to 2022 through May 14.

Counts as supported if: Historical records establish a long-standing, substantial connectionist position opposing symbolic processing, and original contemporary evidence documents unreliable compositional rendering by a leading 2022 image model.

Counts as contradicted if: The historical record fails to contain the asserted substantial opposition, or contemporary evidence establishes reliable handling of the described compositional image tasks. Hybrid research and isolated successful images do not alone contradict the claim.

Retrospective: these criteria were written after the outcome was public.

Current assessment

Was it true for the period described?

Outcome: Supported

The dated evidence matches what the claim committed to, under the criteria frozen before grading.

Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.

Evidence as of
Deadline state when assessed
No horizon
Assessment recorded
Policy version
2026-10-28.7

Rationale

The 1997 and 2019 Stanford Encyclopedia of Philosophy accounts describe many connectionists as rejecting symbolic explanations of cognition, documenting that strand across more than twenty years [e1] [e2]. Both accounts also recognize connectionists who accommodate symbolic processing [e1] [e2]. The DALL·E 2 paper released on April 13, 2022 reports difficulty binding colors to objects in the prompt 'a red cube on top of a blue cube' [e3]. Together, these sources support the historical claim and the contemporary capability claim when 'elude' means unreliable rendering, rather than complete inability [e1] [e2] [e3].

Limitations

  • The historical evidence comprises two versions of one scholarly account, not independent surveys or a count of advocates [e1] [e2].
  • The accounts do not establish unanimous opposition or that the same individuals maintained their positions throughout the period [e1] [e2].
  • The model evidence concerns DALL·E 2 and a GLIDE comparison, not an exhaustive evaluation of every system available by May 14, 2022 [e3].
  • GLIDE's relative advantage does not establish reliable handling of the task, and the reported difficulties do not establish complete inability [e3].

Evidence (3 items)

  • Supports the claim: Stanford Encyclopedia of Philosophy,

    eliminate symbolic processing

    Why it matters here: This dated scholarly account documents an explicit anti-symbolic strand well before the post. It is historical synthesis rather than a primary statement by an advocate.

    Publisher
    Stanford Encyclopedia of Philosophy
    Kind of source
    Secondary summary
    Describes events of
    Public from
  • Supports the claim: Stanford Encyclopedia of Philosophy,

    many connectionists

    Why it matters here: The revised account still describes substantial opposition to explaining cognition through symbolic processing, alongside accommodation and hybrid approaches. Together with the 1997 version, it supports the ordinary historical reading across decades; these are versions of one account, not independent surveys.

    Publisher
    Stanford Encyclopedia of Philosophy
    Kind of source
    Secondary summary
    Describes events of
    Public from
  • Supports the claim: arXiv; OpenAI researchers,

    binding attributes to objects

    Why it matters here: The surrounding passage and Figure 14 examine the red-cube-on-blue-cube prompt and report difficulty assigning the requested colors to the separate objects. This directly supports unreliability on Marcus's example class.

    Publisher
    arXiv; OpenAI researchers
    Kind of source
    Preprint, not peer reviewed
    Describes events of
    Public from

Where it was said (1)

Said once in the sources this register covers.

  • : Marcus on AI, The New Science of Alt Intelligence

    The history missed is the the fact that many neural net enthusiasts, have argued against symbols for decades, and the reality that is missed is the above-documented fact that symbolic descriptions like “red cube on blue cube” still elude the state-of-the-art in 2022.)

    Published
    2022-05-14
    Source observed
    , live page
    Access
    full text, public

History

Assessment versions

Recorded events

  1. Assessment publishedBy this site

    First assessment published after both reviews and the final-text check passed.

    About assessment marcus-the-new-science-of-alt-intelligence-fdddcc-a2.