“In many ways, we have no idea how LLMs actually get used in the real world, e.g. how they get used in housing decisions, loan decisions, crime proceedings etc – but now strong evidence to suspect covert racism to the extent that it is used in such use cases.”
Our reading:
In many ways it is unknown how LLMs are actually used in real-world decisions such as housing, loans, and criminal proceedings, but there is now strong evidence to suspect covert racism to the extent they are used in such cases.
Assessment: Can’t be checked
What happened
Not graded on evidence
Outcome: Can’t be checked
An opinion, a recommendation, or too vague to test. Listed, not scored.
Evidence up to . Checked .
Why
Marcus expresses qualified uncertainty and evidence-based suspicion, not a definite claim that real deployments produced discriminatory outcomes. The original preprint supplies a concrete basis for concern through experimental dialect prejudice. But it cannot settle how much remains unknown or what counts as 'strong evidence to suspect'. Scoring the complete statement would require thresholds he did not supply. Treating the study's findings alone as the claim would also change what is being assessed.
Caveats
- This disposition does not reject the concern or the study's findings.
- No separate claim about documented discrimination in actual deployments is graded.
Evidence (1 item)
Claims like this aren’t scored, so there is no evidence to show.
How we judged it
- Kind of claim
- Observation
- Deadline
- None given. We don’t add one.
Conditions
- to the extent LLMs are used in such use cases (their words)
- in many ways (their words)
- evidence to suspect, not proof (their words)
Where we had to interpret
- The statement expresses a qualified knowledge gap and grounds for concern as of March 5, 2024. It does not assert that discriminatory decisions have been documented in actual deployments.
- The speaker supplies no measurable extent for 'in many ways' or threshold for 'strong evidence to suspect'. Assigning either would add a commitment he did not make.
- The cited study's findings can be checked as context, but grading those findings alone would substitute a narrower claim for the submitted statement.
We don’t grade claims like this. We only check the quote and who said it.
What would settle it
A narrower claim specifying experimental findings, or a claim defining the deployments and observable outcomes at issue, could be assessed separately.
We found 9 claims in this post. See them all.
Full record
Source and dates
Marcus on AI, Covert racism in LLMs, paragraph 13.
- Said
- We read the source
- live page, full text, public
- Evidence up to
- Checked
We read this wording on 2026-10-29. The post may have been edited after 2024-03-05.
Where it was said (1)
Said once in the posts we read.
: Marcus on AI, Covert racism in LLMs
In many ways, we have no idea how LLMs actually get used in the real world, e.g. how they get used in housing decisions, loan decisions, crime proceedings etc – but now strong evidence to suspect covert racism to the extent that it is used in such use cases.
- Where
- paragraph 13
- Published
- 2024-03-05
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Can’t be checked. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
A reviewer said our test wasn’t fair to what was actually said.