“All normal children manage to acquire language in this sense; no existing machine does.”
Our reading, the sentence that was checked:
No existing machine has acquired language in the sense of mapping sentences to meanings and meanings to sentences.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Learning Language is Harder Than You Think, paragraph 40.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-learning-language-is-harder-than-bf4476-a1
We recorded 5 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 2022-07-20. 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
- acquiring language means mapping sentences to meaning and intents to sentences (qualifier, stated by the speaker)
From the speaker
- go from sentences to meaning
- go from intents (meanings) to sentences
- no existing machine does
Our interpretation
- The assessed proposition is the machine claim, not the accompanying generalization about children.
- In context, acquiring language means reliable, general interpretation and expression of meaning comparable in breadth to ordinary human language use.
- A restricted mapping between text and an engineered semantic representation is relevant evidence, but does not alone establish the general competence described.
- Existing means existing on July 20, 2022. With no regular reporting cycle, I use current-year evaluations through that date.
Counts as supported if: Evidence sufficiently broad to establish that contemporary machines did not reliably achieve both general interpretation of sentences and expression of intended meanings.
Counts as contradicted if: At least one contemporary machine reliably demonstrates both abilities across varied, unfamiliar ordinary-language situations.
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
Machines already performed evaluated mappings between sentences and explicit meaning representations in both directions. That matters against a literal reading of no existing machine. But Marcus's context concerns general language competence, which these specialized results do not establish. The evidence neither proves the universal negative nor supplies a qualifying general-purpose counterexample.
Limitations
- Formal meaning representations cover only selected aspects of meaning.
- Choosing the restricted engineering reading would materially change the assessment.
- A finite search cannot establish the capabilities of every existing machine.
Evidence (1 item)
Context only: arXiv; Bai, Chen, and Zhang,
AMR parsing and AMR-to-text generation
Why it matters here: The system performs both directions using Abstract Meaning Representation graphs. This challenges a literal claim that machines could perform no sentence-to-meaning or meaning-to-sentence mapping, but does not establish general humanlike language acquisition.
- Where
- Abstract
- Publisher
- arXiv; Bai, Chen, and Zhang
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
A defensible operational standard for the stated general competence, followed by evaluations of July 2022 systems demonstrating both directions or establishing representative failures against that standard.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Learning Language is Harder Than You Think
All normal children manage to acquire language in this sense; no existing machine does.
- Where
- paragraph 40
- Published
- 2022-07-20
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-learning-language-is-harder-than-bf4476-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-learning-language-is-harder-than-bf4476-a1.