“Language is about a connection between words (or sentence) and the world; the sequences of words that large language models utter lack connection to the external world.”
Our reading, the sentence that was checked:
The word sequences that large language models produce have no connection to the external world, so they lack reference.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Three ideas from linguistics that everyone in AI should know, paragraph 5.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-three-ideas-from-linguistics-that-aa740a-a2
We recorded 26 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 2022-06-22. 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
- the sequences of words that large language models utter
- lack connection to the external world
- Reference
Our interpretation
- The plain reading concerns whether generated language is tied to identifiable entities and situations outside the text, rather than merely whether it sometimes states correct facts.
- The full post distinguishes standalone word-prediction models from models incorporated into externally connected workflows. That distinction must be preserved when assessing tool-assisted systems.
- For this ongoing state, use the current calendar year through publication as the latest reporting window. Research models have no regular quarterly reporting cycle.
Counts as supported if: Documentation and controlled tests establish that the contemporary standalone models lack a mechanism for tying generated references to identifiable external entities or situations.
Counts as contradicted if: A contemporary model within the stated scope demonstrably links its generated references to identifiable external entities or situations, beyond producing coincidentally correct statements.
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
The January 2022 LaMDA paper describes fine-tuning a model to call an external information retrieval system and provide attributed responses [e3]. It reports that 73.2% of responses containing statements about the external world were attributable to known sources [e3]. This demonstrates external information access in a connected system, but does not establish whether standalone models could tie generated references to identifiable external entities or situations [e3].
Limitations
- Source attribution does not by itself establish situated reference or factual correctness [e3].
- The packet contains no controlled test establishing either the presence or absence of situated reference in contemporary standalone models.
- Evidence about a retrieval-connected system cannot alone settle the contract's standalone-model question.
Evidence (1 item)
Context only: arXiv; Thoppilan et al.,
access external knowledge sources.
Why it matters here: LaMDA could consult external sources, and the paper reports source-attributable responses. This challenges an unrestricted claim of no external connection, but does not settle the narrower claim about standalone models and situated reference.
- Publisher
- arXiv; Thoppilan et al.
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Controlled tests of models available during January 1–June 22, 2022, measuring whether generated references track identifiable external entities or situations as context changes, with standalone performance and external-tool contributions measured separately.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Three ideas from linguistics that everyone in AI should know
Language is about a connection between words (or sentence) and the world; the sequences of words that large language models utter lack connection to the external world.
- Where
- paragraph 5
- Published
- 2022-06-22
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-three-ideas-from-linguistics-that-aa740a-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-three-ideas-from-linguistics-that-aa740a-a2.