“OpenAI quietly began doing this to some degree in 2023 with a system called “code interpreter”, in which LLMs (themselves neural networks) call (purely symbolic) Python interpreters. This is literally “putting programs inside”; exactly what Hinton said was a huge mistake.”
Our reading, the sentence that was checked:
In 2023, OpenAI quietly began, to some degree, 'putting programs inside' its systems with Code Interpreter, in which LLMs call purely symbolic Python interpreters. Marcus says this is exactly what Hinton called a huge mistake.
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 75.
- 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-666e04-a1
We recorded 36 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
- 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 degree (qualifier, stated by the speaker)
From the speaker
- OpenAI quietly began doing this to some degree in 2023
- a system called “code interpreter”
- LLMs (themselves neural networks) call (purely symbolic) Python interpreters
- literally “putting programs inside”
- exactly what Hinton said was a huge mistake
Our interpretation
- Assess both the 2023 introduction of Python execution and the claimed correspondence with Hinton's criticism.
- 'Putting programs inside' is read as including a runtime in the overall system workflow, without assuming the interpreter became part of the neural network's weights.
- 'Quietly' has no measurable publicity threshold and is treated as rhetorical.
- The target period is the explicitly named year 2023. Earlier Hinton remarks would provide context for the asserted correspondence.
Counts as supported if: Contemporaneous OpenAI records establish the 2023 Python-execution feature, and an original recording or reliable transcript establishes that Hinton's criticism covered this kind of integration.
Counts as contradicted if: Records disprove the timing or functionality, or Hinton's original remarks clearly exclude the design Marcus says they condemn.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
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
OpenAI's March 2023 announcement verifies Code Interpreter and its Python execution environment. The full claim also says this is exactly what Hinton condemned. The search did not produce a verified original record establishing that correspondence. Supporting the launch component alone is insufficient to support the combined claim, and the missing record is not evidence of contradiction.
Limitations
- The launch and functionality component is supported.
- The Hinton comparison remains unverified.
- System-level integration does not itself establish that symbolic machinery was inserted into the neural architecture.
Evidence (1 item)
Supports the claim: OpenAI,
working Python interpreter
Why it matters here: The announcement confirms that OpenAI introduced a model-accessible Python execution environment in 2023. It supports the timing and functionality component but says nothing about Hinton's criticism.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
An original recording or complete reliable transcript of Hinton's relevant remarks, with enough context to determine whether they cover neural systems calling external symbolic runtimes.
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
OpenAI quietly began doing this to some degree in 2023 with a system called “code interpreter”, in which LLMs (themselves neural networks) call (purely symbolic) Python interpreters. This is literally “putting programs inside”; exactly what Hinton said was a huge mistake.
- Where
- paragraph 75
- Published
- 2025-07-13
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-666e04-a1: 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-666e04-a1.