“As several people include Max Little have noted, the vast majority of the samples of Sora so far largely consist of smoothly panned, slow motion, which fits in some ways with what I am noting above; suggesting the video rely heavily on local patches rather than global understanding.”
Our reading:
The vast majority of Sora samples so far largely consist of smoothly panned slow motion, which suggests the videos rely heavily on local patches rather than global understanding.
Assessment: Unresolved
What happened
Was it true for the period described?
Outcome: Unresolved
We checked, and the evidence so far doesn’t settle it.
Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.
Evidence up to . Checked .
Why
The claim is checkable, but the located records do not establish the denominator or the share of qualifying videos. The preview's prompts cannot establish actual playback motion. Neither selected slow examples nor isolated active examples settle a vast-majority claim. The tentative explanation about local patches also remains unproven.
Caveats
- The current gallery is not a verified snapshot of everything public at the post's publication time.
- No numerical threshold should be invented after inspecting the examples.
Evidence (1 item)
Context only: OpenAI,
A cartoon kangaroo disco dances.
Why it matters: The gallery includes prompts requesting active subjects. A prompt does not establish the generated video's speed or camera motion, and this example supplies no prevalence estimate.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- About events on
- Published
How we judged it
- Kind of claim
- Observation
- Deadline
- None given. We don’t add one.
Conditions
- samples so far (their words)
- suggesting (their words)
- fits in some ways (their words)
Where we had to interpret
- The checkable commitment is the prevalence of smooth, slow-moving shots among unique Sora samples publicly available by February 17, 2024.
- Vast majority means a clearly overwhelming share, not merely more than half. No precise percentage is supplied or imposed.
- Each sample should be classified by its predominant visible motion, distinguishing camera movement from subject movement.
- The local-patches explanation is a tentative inference. Motion prevalence alone cannot establish the model's internal understanding.
- Samples so far includes public launch material and publicly shared demonstrations, rather than only examples selected by the evaluator.
Right if: A dated, sufficiently complete inventory and reproducible motion review show that a clearly overwhelming share of the eligible samples predominantly use smooth pans and slow motion.
Wrong if: The same review shows that such samples are clearly not an overwhelming share. A few fast or multi-shot examples alone do not contradict a majority claim.
We wrote this test after the outcome was known.
What would settle it
A dated archive of the eligible videos and an independently reproducible count of predominant camera and subject motion, with sensitivity to ambiguous classifications.
We found 12 claims in this post. See them all.
Full record
Source and dates
Marcus on AI, Sora Can’t Handle the Truth, paragraph 22.
- 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 2024-02-17.
Where it was said (1)
Said once in the posts we read.
: Marcus on AI, Sora Can’t Handle the Truth
As several people include Max Little have noted, the vast majority of the samples of Sora so far largely consist of smoothly panned, slow motion, which fits in some ways with what I am noting above; suggesting the video rely heavily on local patches rather than global understanding.
- Where
- paragraph 22
- Published
- 2024-02-17
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.