“The technology we have now is built on autocompletion, not factuality.”
Our reading, the sentence that was checked:
Current generative AI is built on autocompletion rather than factuality.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What if Generative AI turned out to be a Dud?, paragraph 19.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-if-generative-ai-turned-out-f20765-a1
We recorded 23 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 2023-08-13. 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
- technology we have now
- built on autocompletion
- not factuality
Our interpretation
- In context, technology refers to contemporary general-purpose language models.
- The plain reading of built on is their foundational training objective: predicting text continuations rather than verifying each assertion against facts.
- This does not mean that factuality-oriented fine-tuning, retrieval, or external verification were absent.
- With no common reporting cycle, the current-system observation window is calendar 2023 through publication.
Counts as supported if: Primary technical documentation identifies predictive text completion as the foundational objective, while factuality improvements or verification are additional mechanisms.
Counts as contradicted if: Primary documentation establishes factual verification as the foundational mechanism of the contemporary systems instead of predictive text completion.
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
GPT-4’s technical report expressly describes next-token prediction as its pretraining objective. That supports the ordinary foundation reading of autocompletion. The qualification matters: the same report describes factuality improvements from post-training, and RARR demonstrates external grounding and revision. A stronger claim that contemporary systems contained no factuality-oriented components would not be supported.
Limitations
- The outcome depends on reading built on as the foundation rather than every component of a deployed system.
- Predicting text does not imply an inability to represent facts.
- The conclusion concerns language models, not all generative architectures.
Evidence (2 items)
Supports the claim: OpenAI, arXiv,
predict the next token in a document
Why it matters here: The report explicitly identifies predictive completion as GPT-4’s base objective. It separately reports factuality gains from post-training, supporting the foundation reading while qualifying a literal denial of factuality work.
- Publisher
- OpenAI, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Luyu Gao and colleagues, arXiv,
standard web search
Why it matters here: RARR retrieves evidence and revises unsupported generated content, showing that factual grounding could augment an underlying language model before Marcus wrote. The method predates the observation window and is contextual evidence against an overly broad reading.
- Publisher
- Luyu Gao and colleagues, arXiv
- 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, What if Generative AI turned out to be a Dud?
The technology we have now is built on autocompletion, not factuality.
- Where
- paragraph 19
- Published
- 2023-08-13
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-if-generative-ai-turned-out-f20765-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-what-if-generative-ai-turned-out-f20765-a1.