“We can of course expect related problems of discomprehension in efforts to use these systems to address copyright issues, as I wrote earlier this week.”
Our reading, the sentence that was checked:
Efforts to use generative AI systems to address copyright issues will run into problems of discomprehension.
Assessment: Supported Settled before its deadline on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Where’s Waldo? The Elephant in the Room, paragraph 10.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-wheres-waldo-the-elephant-in-the-d71263-a1
We recorded 9 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 2024-01-25. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- Settled before its deadline
- Deadline state on 2026-10-29
- Settled before its deadline (the release’s reference date)
Conditions
- when used to address copyright issues (qualifier, stated by the speaker)
From the speaker
- expect related problems of discomprehension
- use these systems to address copyright issues
- GenAI used for source attribution
Our interpretation
- The condition includes actual research evaluations of generative models used for copyright screening, as well as deployments.
- A substantive misidentification of protected characters or attribution of material to the wrong source counts as the predicted problem.
- This is a prediction that failures will occur, not that every attempt will fail or that copyright problems are impossible to mitigate.
- A model's disagreement with a benchmark label is evidence about screening performance, not a judicial determination of infringement.
- A dated public release documenting a completed qualifying evaluation can resolve this occurrence forecast.
Counts as supported if: After the post, a documented copyright-screening or source-attribution use of generative AI exhibits substantive identification or attribution errors.
Counts as contradicted if: A comprehensive record establishes that qualifying uses avoid the predicted errors throughout an independently specified assessment interval. Individual successful uses do not contradict this occurrence forecast.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
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
- Settled before its deadline
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
A February 2025 empirical report documents generative vision-language models being evaluated for copyright screening and making substantive false-positive classifications. This satisfies both the condition and the predicted occurrence under the stated reading. It does not show that every copyright tool fails, that mitigation is impossible, or that the benchmark labels establish legal infringement.
Limitations
- The decisive study is a preprint.
- Support depends on including research evaluations within efforts to address copyright issues.
- The assessment concerns screening mistakes against the study's labels, not binding legal judgments.
- The resolution date is the public release of the completed evaluation.
Evidence (3 items)
Supports the claim: arXiv; Xu and colleagues,
LVLMs
Why it matters here: The public report documents a completed evaluation using generative vision-language models for copyright detection. The event date is the release of that empirical report; individual testing dates are not specified.
- Where
- Abstract
- Publisher
- arXiv; Xu and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Supports the claim: arXiv; Xu and colleagues,
misjudgment
Why it matters here: The released evaluation documents mistaken copyright classifications and analyzes reliance on isolated character features. This meets the operational reading of substantive screening errors, subject to the authors' acknowledged limits on legal ground truth.
- Publisher
- arXiv; Xu and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Tow Center for Digital Journalism, Columbia Journalism Review,
cite the wrong article
Why it matters here: Direct source-identification tests found erroneous attribution, including problems involving publisher partners. This corroborates the attribution concern, but news retrieval is broader than copyright screening and is not the decisive evidence.
- Publisher
- Tow Center for Digital Journalism, Columbia Journalism Review
- Kind of source
- Independent measurement
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Where’s Waldo? The Elephant in the Room
We can of course expect related problems of discomprehension in efforts to use these systems to address copyright issues, as I wrote earlier this week.
- Where
- paragraph 10
- Published
- 2024-01-25
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-wheres-waldo-the-elephant-in-the-d71263-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-wheres-waldo-the-elephant-in-the-d71263-a1.