“Sora is never really getting it, even with mountains and mountains of data.”
Our reading, the sentence that was checked:
Sora will never really grasp object permanence, even with vast amounts of data.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Sora still appears to have trouble with physics, paragraph 5.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-sora-still-appears-to-have-trouble-ed695f-a1
We recorded 4 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 2024-12-10. 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
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
Conditions
- even with mountains of data (qualifier, stated by the speaker)
From the speaker
- Sora is never really getting it
- even with mountains and mountains of data
- object permanence
Our interpretation
- In context, it refers to object permanence and continuity across occlusion or camera movement.
- The plain reading is that later Sora versions will not reliably preserve objects and their identities when temporarily hidden.
- Even with mountains of data is a concession about extensive training, not a specified future antecedent. No dataset size, increase or training intervention is given.
- Reliability does not require every generated frame to be perfect.
Counts as supported if: Decisive evidence would need to establish an enduring inability to maintain object permanence despite extensive training. Repeated failures in available models are provisional evidence only.
Counts as contradicted if: A later Sora version reliably preserves object existence and identity across varied, independently tested occlusions and camera movements.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
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 deadline stated
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Later testing that includes Sora 2 Pro finds both coherence successes and substantial failures under interrupted observation. It neither proves permanent incapacity nor establishes reliable object permanence across varied situations. The phrase about extensive data supplies no measurable antecedent or deadline.
Limitations
- The cited aggregate sentence does not give a Sora-only success rate.
- Automated coherence evaluation covers more than object permanence.
- Training-data effects are not isolated.
Evidence (1 item)
Context only: Ziqi Ma, Mengzhan Liufu and Georgia Gkioxari / arXiv,
The coherence verifier captures these errors, with state-of-the-art closed-source models achieving only about a 60% success rate.
Why it matters here: The experiments include Sora 2 Pro, and the coherence test covers unexplained object loss, appearance and displacement. The reported successes and failures bear on current capability, but this broader coherence measure cannot settle an unlimited forecast about object permanence.
- Publisher
- Ziqi Ma, Mengzhan Liufu and Georgia Gkioxari / arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Independent, reproducible tests showing reliable object permanence across varied occlusions and camera movements in a later Sora version would contradict the forecast. Continued failures alone cannot establish never.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Sora still appears to have trouble with physics
Sora is never really getting it, even with mountains and mountains of data.
- Where
- paragraph 5
- Published
- 2024-12-10
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-sora-still-appears-to-have-trouble-ed695f-a1: 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-sora-still-appears-to-have-trouble-ed695f-a1.