“For another, chatGPT, just released by OpenAI, is more or less just as capable of producing similar nonsense, such as instant essays on adding wood chips to breakfast cereal.”
Our reading, the sentence that was checked:
ChatGPT is about as capable as Galactica of producing nonsense misinformation.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, AI's Jurassic Park moment, paragraph 12.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-ais-jurassic-park-moment-ace0c5-a2
We recorded 15 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 2022-12-10. 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
- chatGPT, just released by OpenAI
- more or less just as capable of producing similar nonsense
- instant essays on adding wood chips to breakfast cereal
Our interpretation
- I read 'more or less just as capable' as a rough comparison, not a measured one: like Galactica, the newly released ChatGPT readily produced fluent, confident falsehoods, and its safeguards could be got around.
- Target period runs from ChatGPT's release (2022-11-30) to the post date (2022-12-10).
Counts as supported if: Documented reports from the period that ChatGPT readily produced plausible-sounding but false content, comparable to what Galactica was criticized for.
Counts as contradicted if: Evidence from the period that ChatGPT reliably refused or avoided producing confident false content, so it was clearly less prone to it than Galactica.
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
On 2022-12-10, Marcus wrote that the newly released ChatGPT was "more or less just as capable" as Meta's Galactica of producing similar nonsense. On 2022-12-05, The Verge reported that users had noted ChatGPT's persistent tendency to give plausible but false answers, including wrong biographical details stated with confidence (e1). The same day, The Register quoted Stack Overflow saying ChatGPT's answers had a high rate of being incorrect but typically looked good and were easy to produce (e2). Vice reported that within hours of launch, users found ways around ChatGPT's content filters (e3). MIT Technology Review reported in November 2022 that Galactica made up fake papers and wrote articles about bears in space (e4). Taken together, the reports from that period show ChatGPT producing fluent false content of the same kind reported for Galactica. That fits the contract's rough, qualitative reading. The evidence packet found no test that compared the two models on the same prompts.
Limitations
- The packet found no head-to-head test of ChatGPT and Galactica, so 'more or less just as capable' is judged qualitatively, not measured.
- His specific example, essays on adding wood chips to breakfast cereal, was not independently verified.
- The reports describe user observations and a platform policy statement, not systematic error rates.
Evidence (4 items)
Supports the claim: The Verge,
others have noted its persistent tendency to generate plausible but false responses. Ask the bot to write a biography of a public figure, for example, and it may well insert incorrect biographical data with complete confidence.
Why it matters here: In its first week ChatGPT was reported to confidently produce plausible but false content, the same failure Galactica was pulled for.
- Where
- paragraph 7
- Publisher
- The Verge
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: The Register,
"While the answers which ChatGPT produces have a high rate of being incorrect, they typically look like they might be good and the answers are very easy to produce," Stack Overflow said.
Why it matters here: Stack Overflow, from direct experience, found ChatGPT output often wrong yet convincing and cheap to produce in volume.
- Where
- paragraph 7
- Publisher
- The Register
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: Vice (Motherboard),
Within hours of the chatbot’s demo appearing online, social media users began discovering ways to circumvent OpenAI’s content filters. While the bot initially refuses to answer questions on how to commit crimes, it will comply if the query is phrased in certain ways.
Why it matters here: ChatGPT had refusal filters, but they were easy to get around, so the filters did not make it clearly less capable of producing harmful output.
- Where
- paragraph 6
- Publisher
- Vice (Motherboard)
- Kind of source
- News report
- Describes events of
- Public from
Context only: MIT Technology Review,
People found that it made up fake papers (sometimes attributing them to real authors), and generated wiki articles about the history of bears in space
Why it matters here: Shows the kind of nonsense Galactica produced, which is the point of comparison.
- Where
- paragraph 11
- Publisher
- MIT Technology Review
- Kind of source
- News report
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, AI's Jurassic Park moment
For another, chatGPT, just released by OpenAI, is more or less just as capable of producing similar nonsense, such as instant essays on adding wood chips to breakfast cereal.
- Where
- paragraph 12
- Published
- 2022-12-10
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-ais-jurassic-park-moment-ace0c5-a2: 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.