“And the effects of all that IP theft on artists and writers, and potential negative effects more broadly across the employment market (though, mercifully, these have been more modest than some people expected).”
Our reading, the sentence that was checked:
Generative AI's negative effects on the employment market have so far been more modest than some expected, while its IP theft has harmed artists and writers.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Will Generative AI do more harm than good?, paragraph 2.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-will-generative-ai-do-more-harm-than-d577cd-a2
We recorded 5 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-10-24. 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 effects of all that IP theft on artists and writers
- potential negative effects more broadly across the employment market
- these have been more modest than some people expected
Our interpretation
- The plain reading is that some artists and writers have suffered harm from unauthorized appropriation of their work, while broader employment harms have been smaller than some earlier expectations.
- The parenthetical refers to broader employment effects. It does not deny substantial harm to particular occupations.
- For this ongoing observation, use September 2026, the latest completed calendar month before the post, to assess the position reached so far. Earlier events provide historical context.
- An expectation must concern employment effects by the assessment period. A prediction about later years cannot be treated as having failed already.
- Read IP theft as unauthorized appropriation, without assuming that every unlicensed training use is legally infringing.
- Marcus supplies no geography. Country-specific findings can inform the assessment but cannot establish a worldwide result by themselves.
Counts as supported if: Evidence for the assessment period establishes both components: broader AI-related employment harms smaller than identifiable earlier expectations for the same population and elapsed period, and documented harm to artists and writers linked to unauthorized appropriation of their work.
Counts as contradicted if: Comparable evidence shows employment harms meeting or exceeding the identified earlier expectations, or reliable evidence refutes the alleged appropriation-related creator harms. Benefits elsewhere or lawful uses in particular cases do not alone refute the claim.
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
Yale's October 14, 2026 assessment reported no connection between AI usage and changes in employment or unemployment, but its readable text does not identify the observation cutoff [e1]. Forrester's January 13, 2026 jobs forecast runs through 2030; it does not supply a comparable expectation due by September 2026 [e2]. ISM's January 30, 2026 summary reported creator losses and scraping without consent, but does not establish September 2026 losses caused by that copying [e3]. The June 23, 2025 court order found the training use at issue fair use and left pirated library copies and resulting damages for trial; it does not establish creator harm for September 2026 [e4]. These sources settle neither the comparison with earlier employment expectations nor the claimed appropriation-related creator harms for the assessment period [e1][e2][e3][e4].
Limitations
- No verified earlier employment expectation matches measured September 2026 outcomes in population, measure and elapsed period [e1][e2][e5].
- The available creator summary does not establish sampling methods, observation dates or a causal link between unauthorized copying and reported losses [e3].
- The court's findings concern particular copying and training uses; they do not refute creator harms generally [e4].
- The US employment findings and UK creator summary do not establish a worldwide result [e2][e3][e5].
Evidence (5 items)
Context only: The Budget Lab at Yale,
no connection to changes in employment or unemployment.
Why it matters here: This dated research assessment is consistent with limited broad disruption. Its readable text supplies neither an underlying observation cutoff nor an earlier expectation for comparison; the event date identifies the assessment, not a verified September measurement.
- Publisher
- The Budget Lab at Yale
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: Forrester,
Financially driven layoffs
Why it matters here: The disclosure warns against attributing announced layoffs automatically to AI and presents a forecast through 2030. That future endpoint does not provide a comparable expectation already due in September 2026.
- Publisher
- Forrester
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Independent Society of Musicians,
the UK’s creators.
Why it matters here: A report publisher describes creator evidence of lost work and unauthorized scraping. This is a January summary by an advocacy organization; the inaccessible underlying report prevents verification of observation dates, sampling, and the link between copying and losses.
- Publisher
- Independent Society of Musicians
- Kind of source
- Secondary summary
- Describes events of
- Public from
Context only: U.S. District Court, Northern District of California; copy hosted by Hunton,
the training use
Why it matters here: The court distinguished training use from downloaded pirated copies retained in a central library, which it did not justify as fair use. This limits a blanket legal reading of IP theft, but neither disproves particular creator harms nor measures their September 2026 extent.
- Where
- Conclusion, line 23
- Publisher
- U.S. District Court, Northern District of California; copy hosted by Hunton
- Kind of source
- Regulatory filing
- Describes events of
- Public from
Context only: Goldman Sachs Research,
younger, less-experienced
Why it matters here: The research summary describes modest aggregate effects alongside concentrated harm to younger workers. It concerns the preceding year without an exact observation cutoff, so it is historical context rather than September outcome evidence.
- Publisher
- Goldman Sachs Research
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Employment outcomes through September 2026 compared with an identifiable earlier expectation for the same population, measure and period, plus evidence establishing or refuting harm to artists and writers caused by unauthorized appropriation by that period. Both components need a defined geographical scope.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Will Generative AI do more harm than good?
And the effects of all that IP theft on artists and writers, and potential negative effects more broadly across the employment market (though, mercifully, these have been more modest than some people expected).
- Where
- paragraph 2
- Published
- 2026-10-24
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-will-generative-ai-do-more-harm-than-d577cd-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-will-generative-ai-do-more-harm-than-d577cd-a2.