“To some extent newer models are doing that.”
Our reading, the sentence that was checked:
Newer models such as DeepSeek R1, with a rule-based reward system, partly use neurosymbolic techniques.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Five ways in which the last 3 months — and especially the DeepSeek era — have vindicated “Deep learning is hitting a wall", paragraph 5.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-five-ways-in-which-the-last-3-months-f8bc64-a1
We recorded 8 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 2025-02-09. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- 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)
Conditions
- to some extent (qualifier, stated by the speaker)
From the speaker
- combining neural networks with classical symbolic techniques such as rules
- To some extent newer models are doing that
- DeepSeek’s R1 model
- rule-based reward system
- verifying some classes of answers
Our interpretation
- The contextual period is the three months preceding the post.
- The speaker expressly includes programmed training-time answer checks within his modest description of combining neural networks and rules.
- The claim does not require a symbolic inference engine inside the deployed neural model.
- DeepSeek R1 is the concrete example to test; no inference about o1's undisclosed implementation is required.
Counts as supported if: The contemporaneous R1 technical documentation describes a neural model trained using explicit programmed correctness or formatting checks.
Counts as contradicted if: The documentation establishes that the attributed checks were absent or entirely learned, with no programmed verification rules in the described training procedure.
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
DeepSeek's original technical report documents neural training with programmed answer checks and explicitly applies rule-based reasoning rewards to R1. This supports Marcus's qualified claim under the definition supplied in his paragraph. It does not establish that R1 performs symbolic inference internally or that the field uniformly classifies such reward systems as neurosymbolic.
Limitations
- The outcome depends on the broad, training-inclusive meaning supplied by the speaker's example.
- The source is the developer's technical report, not an independent audit.
- The evidence does not establish any claim about o1 or AlphaFold.
Evidence (1 item)
Supports the claim: DeepSeek-AI, arXiv,
predefined test cases
Why it matters here: Section 2.2.2 describes programmed accuracy and formatting rewards for R1-Zero, including code tests. Section 2.3.4 explicitly applies its rule-based reasoning reward methodology to R1, supporting the limited training-time combination described by Marcus.
- Publisher
- DeepSeek-AI, arXiv
- 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, Five ways in which the last 3 months — and especially the DeepSeek era — have vindicated “Deep learning is hitting a wall"
To some extent newer models are doing that.
- Where
- paragraph 5
- Published
- 2025-02-09
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-five-ways-in-which-the-last-3-months-f8bc64-a1: Current. Outcome: Supported. 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-five-ways-in-which-the-last-3-months-f8bc64-a1.