“Human mathematicians had to write problem-specific code for each separate mathematical problem.”
Our reading, the sentence that was checked:
FunSearch required human mathematicians to write problem-specific code for each separate problem.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Sorry, but FunSearch probably isn’t a milestone in scientific discovery, paragraph 11.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-sorry-but-funsearch-probably-isnt-63781a-a1
We recorded 7 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 2023-12-19. 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)
From the speaker
- Human mathematicians had to write
- problem-specific code
- for each separate mathematical problem
Our interpretation
- The claim concerns the setup required for different problem families in the original FunSearch work.
- Human mathematicians is read as the human researchers specifying the mathematical tasks, not as a requirement for particular professional credentials.
- Each separate problem means a different task, such as cap-set construction versus bin packing, rather than every input instance or dimension.
- Required problem-specific code is distinguished from an optional program skeleton.
Counts as supported if: The original method requires users to supply a coded task specification and initial program, with task-specific evaluation code for the different problems studied.
Counts as contradicted if: The original method takes each new mathematical task without human-supplied task-specific code and itself constructs the necessary specification and evaluation 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
The original method required a coded evaluator and initial program for the task. Its examples use different specifications for cap sets and bin packing. This supports the ordinary reading that human researchers supplied problem-specific code. It does not establish a credential requirement or that fresh code was needed for every instance.
Limitations
- The outcome depends on interpreting mathematicians as human task designers and separate problems as different task families.
- The paper treats a skeleton as beneficial rather than strictly necessary in principle.
Evidence (1 item)
Supports the claim: Nature / Google DeepMind author manuscript,
candidate solutions
Why it matters here: The specification section requires a coded evaluator and an initial program. Figure 2 supplies different evaluation and surrounding code for cap sets and bin packing, supporting the reading that humans supplied task-specific code.
- Publisher
- Nature / Google DeepMind author manuscript
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Sorry, but FunSearch probably isn’t a milestone in scientific discovery
Human mathematicians had to write problem-specific code for each separate mathematical problem.
- Where
- paragraph 11
- Published
- 2023-12-19
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-sorry-but-funsearch-probably-isnt-63781a-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-sorry-but-funsearch-probably-isnt-63781a-a1.