“Very little of what has been tried has been discussed explicitly, and because the companies are often quite closed about what they are doing, the public science of neurosymbolic AI is greatly impoverished.”
Our reading, the sentence that was checked:
Company secrecy has greatly impoverished public science on neurosymbolic AI.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, How o3 and Grok 4 Accidentally Vindicated Neurosymbolic AI, paragraph 119.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-how-o3-and-grok-4-accidentally-vindicated-46586b-a2
We recorded 47 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-07-13. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Causal claim
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
From the speaker
- Very little of what has been tried has been discussed explicitly
- companies are often quite closed
- because
- the public science of neurosymbolic AI is greatly impoverished
Our interpretation
- The plain reading is that withheld company methods substantially reduced publicly available, reproducible knowledge of neurosymbolic AI by the post date.
- Opacity alone does not establish the asserted size or cause of the research deficit.
- A public paper is evidence of disclosure, but does not establish that all attempted methods or necessary replication details were disclosed.
- I do not invent numerical thresholds for very little or greatly.
Counts as supported if: Evidence identifies withheld neurosymbolic methods and shows that their withholding substantially obstructed replication, comparison, or public research progress.
Counts as contradicted if: Broad disclosure and replication evidence establishes that company withholding did not materially restrict public neurosymbolic knowledge.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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 2024 Foundation Model Transparency Index documented continuing opacity alongside an increase in average transparency scores from 37 in 2023 to 58 in 2024 [e1]. It assessed foundation-model developers broadly, without measuring losses to neurosymbolic research [e1]. The February 5, 2025 AlphaGeometry2 paper disclosed changes to its language model, symbolic engine, formal language, and search method [e2]. These sources establish neither the fraction of company neurosymbolic methods withheld nor substantial research losses caused by withholding, leaving the causal claim unresolved [e1, e2].
Limitations
- The transparency audit does not isolate neurosymbolic methods or their research impact [e1].
- One methodological disclosure cannot establish the overall disclosure rate or full reproducibility [e2].
- The packet contains no comparison separating effects of withholding from technical difficulty or other research constraints [e1, e2].
Evidence (2 items)
Context only: Bommasani et al., TMLR; arXiv version,
extremely opaque.
Why it matters here: The audit documents substantial opacity but also an improvement in average transparency from 37 to 58 out of 100. It concerns foundation-model transparency generally and does not measure lost neurosymbolic research or establish that secrecy caused such a loss.
- Publisher
- Bommasani et al., TMLR; arXiv version
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
Context only: Google DeepMind researchers, arXiv,
a symbolic engine
Why it matters here: The AlphaGeometry2 paper explicitly describes its neural-symbolic predecessor and details changes to the language model, symbolic engine, formal language, and search. This is material evidence of company disclosure, although one disclosed system cannot establish the overall disclosure rate or eliminate research losses elsewhere.
- Publisher
- Google DeepMind researchers, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
An inventory of disclosed and withheld company neurosymbolic methods through July 13, 2025, with replication and research-impact evidence showing whether withholding substantially restricted public progress after accounting for other causes.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, How o3 and Grok 4 Accidentally Vindicated Neurosymbolic AI
Very little of what has been tried has been discussed explicitly, and because the companies are often quite closed about what they are doing, the public science of neurosymbolic AI is greatly impoverished.
- Where
- paragraph 119
- Published
- 2025-07-13
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-46586b-a2: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-how-o3-and-grok-4-accidentally-vindicated-46586b-a2.