“A good system should give the user a manifest of sources; current systems don’t.”
Our reading, the sentence that was checked:
Current generative AI systems do not give users a manifest of sources.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Things are about to get a lot worse for Generative AI, paragraph 28.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-things-are-about-to-get-a-lot-worse-679273-2-a1
We recorded 14 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 2023-12-29. 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
- a manifest of sources
- current systems don’t
Our interpretation
- Assess the factual second clause; the statement about what a good system should do is normative.
- In the surrounding discussion, a manifest means a reliable account of the underlying source materials behind generated outputs, including training sources.
- The ordinary contextual scope is mainstream foundation-model systems such as DALL-E and ChatGPT. A bibliography of retrieved pages is not necessarily a training-source manifest.
- Use December 2023 as the latest contemporaneous capability snapshot before the post. This is a month-level description of the state at publication, not an assessment of later product changes.
Counts as supported if: Evidence addressing December 2023 shows that mainstream systems did not provide users with reliable output-specific manifests of underlying training sources.
Counts as contradicted if: A mainstream system available at the time demonstrably provided such a manifest to users.
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 December 2023 research account supports the contextual claim that foundation-model systems lacked the provenance support needed for reliable training-source manifests. Existing search citations are an important qualification: the sentence would be too broad if read as denying all source lists. The proposed support therefore applies to underlying-source manifests, as suggested by the surrounding discussion of training and provenance.
Limitations
- The outcome depends on the contextual meaning of “manifest”.
- The decisive source is a technical synthesis published later and explicitly describing its December 2023 writing period.
- It does not establish the separate causal claim that attribution was architecturally impossible.
Evidence (2 items)
Supports the claim: An MIT Exploration of Generative AI; author-hosted copy,
current foundation models.
Why it matters here: The authors describe the state at their December 2023 writing date and identify the missing provenance relationships needed for attribution. This supports the contextual reading about underlying source manifests, although it is a research synthesis rather than an exhaustive product audit.
- Publisher
- An MIT Exploration of Generative AI; author-hosted copy
- Kind of source
- Secondary summary
- Describes events of
- Public from
Context only: arXiv; Nelson F. Liu, Tianyi Zhang, and Percy Liang,
in-line citations.
Why it matters here: Generative search engines already provided citations before the post. This is material counterevidence to a reading that denies any source lists, but citations do not establish complete training-source provenance.
- Publisher
- arXiv; Nelson F. Liu, Tianyi Zhang, and Percy Liang
- 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, Things are about to get a lot worse for Generative AI
A good system should give the user a manifest of sources; current systems don’t.
- Where
- paragraph 28
- Published
- 2023-12-29
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-things-are-about-to-get-a-lot-worse-679273-2-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-things-are-about-to-get-a-lot-worse-679273-2-a1.