MIT Technology Review · Aug 27, 2024
MIT Technology Review’s guide to judging humanoid-robot videos
Markdown · JSON · Original source
Published August 27, 2024, MIT Technology Review’s unsigned newsletter article uses roboticist Ken Goldberg’s advice to explain why polished humanoid-robot videos can mislead viewers about what robots can do. The article says videos matter because people cannot casually try a humanoid robot, making them a key marketing tool for startups and a way for academics to demonstrate progress when seeking lab funding. To judge a video, it recommends asking whether a person is tele-operating the robot, how many takes were needed, and whether editing speeds up footage in a way that hides limitations. It also cautions that a hand gripping a box may be squeezing it sideways, that impressive movement is often filmed in pristine environments, and that boxes may be empty. Goldberg says usable robot hands require complex engineering, and their apparent dexterity may exceed what they can actually do. The article’s point is not that the videos show nothing: it calls some feats impressive, while urging viewers to distinguish the filmed performance from what a robot can manage in ordinary work settings. It leaves open how much humanoid robots may cost.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
MIT Technology Review · Article
Source details
Original source date: Aug 27, 2024 (2024-08-27T09:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-17T10:03:23.118Z. First observed: 2026-10-29T23:22:26.722Z. Retrieved: 2026-10-29T23:23:50.572Z.
Referenced by
Recorded citations in public readings. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 29, 2026; originally published August 27, 2024. The article does not provide evidence of how often the filmed tasks succeed or how well robots perform in ordinary work settings. The article says the eventual cost of humanoid robots is an open question.
Open question
What evidence would show that a humanoid robot can perform these tasks reliably in the less controlled settings where it is expected to work?
Source-stated
The article says humanoid-robot videos serve as marketing for startups and can help academics seeking lab funding demonstrate progress.
a good humanoid video is the best marketing tool available
Source-stated
Goldberg says viewers cannot tell from a video whether a robot is tele-operated unless the company discloses that information.
you can’t tell unless the company discloses it in the video
Source-stated
The article cautions that a robot shown holding a box may be squeezing it horizontally rather than lifting it with its fingers.
the robots are actually just squeezing the box horizontally, with the force coming from the shoulder.
Source-stated
The article says robot feats are generally shown in pristine settings unlike the warehouses and factories where the robots are expected to work.
The robots generally show off their (admittedly impressive) feats in pristine environments
Source-stated
Goldberg says a robot that could outrun a human is probably near, presenting this as his prediction.
“A robot that could outrun a human is probably right around the corner,” Goldberg says.
Source-stated
The article says humanoid robots’ carrying capacity varies dramatically when their arms are outstretched.
their carrying capacity varies dramatically as their arms are outstretched.
Escape Meridian interpretation
Escape Meridian reads the article as distinguishing impressive filmed performances from demonstrated ability in ordinary work settings.
not quite like the warehouses, factories, and other places where they will purportedly work alongside humans.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: 415c3f4d-3347-27d7-b63d-fdee1c06d7f0.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-17T09:59:12.565Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 6a35b880dcb5455f29f02a3f0cb839550ef9be8e68141c41cb3d2ffb9a81ae14. Source text hash: a350205a937525380215f00aba40d03977d076e5e0ebdc0cc140b9b47073d668.
Revision and source-check history
Version 1 · 2026-11-17T10:03:23.118Z
MIT Technology Review’s guide to judging humanoid-robot videos
Published August 27, 2024, MIT Technology Review’s unsigned newsletter article uses roboticist Ken Goldberg’s advice to explain why polished humanoid-robot videos can mislead viewers about what robots can do. The article says videos matter because people cannot casually try a humanoid robot, making them a key marketing tool for startups and a way for academics to demonstrate progress when seeking lab funding. To judge a video, it recommends asking whether a person is tele-operating the robot, how many takes were needed, and whether editing speeds up footage in a way that hides limitations. It also cautions that a hand gripping a box may be squeezing it sideways, that impressive movement is often filmed in pristine environments, and that boxes may be empty. Goldberg says usable robot hands require complex engineering, and their apparent dexterity may exceed what they can actually do. The article’s point is not that the videos show nothing: it calls some feats impressive, while urging viewers to distinguish the filmed performance from what a robot can manage in ordinary work settings. It leaves open how much humanoid robots may cost.
Original source
MIT Technology Review · Article
Source details
Original source date: Aug 27, 2024 (2024-08-27T09:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-17T10:03:23.118Z. First observed: 2026-10-29T23:22:26.722Z. Retrieved: 2026-10-29T23:23:50.572Z.
Limits
Text as retrieved on October 29, 2026; originally published August 27, 2024. The article does not provide evidence of how often the filmed tasks succeed or how well robots perform in ordinary work settings. The article says the eventual cost of humanoid robots is an open question.
Open question
What evidence would show that a humanoid robot can perform these tasks reliably in the less controlled settings where it is expected to work?
Source-stated
The article says humanoid-robot videos serve as marketing for startups and can help academics seeking lab funding demonstrate progress.
a good humanoid video is the best marketing tool available
Source-stated
Goldberg says viewers cannot tell from a video whether a robot is tele-operated unless the company discloses that information.
you can’t tell unless the company discloses it in the video
Source-stated
The article cautions that a robot shown holding a box may be squeezing it horizontally rather than lifting it with its fingers.
the robots are actually just squeezing the box horizontally, with the force coming from the shoulder.
Source-stated
The article says robot feats are generally shown in pristine settings unlike the warehouses and factories where the robots are expected to work.
The robots generally show off their (admittedly impressive) feats in pristine environments
Source-stated
Goldberg says a robot that could outrun a human is probably near, presenting this as his prediction.
“A robot that could outrun a human is probably right around the corner,” Goldberg says.
Source-stated
The article says humanoid robots’ carrying capacity varies dramatically when their arms are outstretched.
their carrying capacity varies dramatically as their arms are outstretched.
Escape Meridian interpretation
Escape Meridian reads the article as distinguishing impressive filmed performances from demonstrated ability in ordinary work settings.
not quite like the warehouses, factories, and other places where they will purportedly work alongside humans.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: 415c3f4d-3347-27d7-b63d-fdee1c06d7f0.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-17T09:59:12.565Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 6a35b880dcb5455f29f02a3f0cb839550ef9be8e68141c41cb3d2ffb9a81ae14. Source text hash: a350205a937525380215f00aba40d03977d076e5e0ebdc0cc140b9b47073d668.
uncertain · 2026-11-17T09:59:12.565Z · anthropic/claude-opus-5.5
Publication: 415c3f4d-3347-27d7-b63d-fdee1c06d7f0
attribution — uncertain: The reading correctly attributes the advice to MIT Technology Review's newsletter piece and Ken Goldberg. However, it calls the article "unsigned." The captured text is written in the first person by an unnamed author, and no byline appears in the evidence. That does not establish that the original was unsigned; the byline may simply be missing from the capture. Smallest repair: remove "unsigned," or say the author is not named in the retrieved text.
I’ve watched dozens of them this year, and I confess I frequently oscillate between being impressed, scared, and bored.
summary — pass: The title, summary and question accurately present the skeptical guide: tele-operation, number of takes, sped-up footage, hands, pristine environments and empty boxes. The summary explains tele-operation in plain words, keeps the author's acknowledgment that some feats are impressive, and states the cost question the article leaves open. A non-specialist can follow it. The question follows from the source's point about pristine environments.
date — pass: August 27, 2024 matches the UTC calendar day of evidence.originalPublished. The retrieval note is clearly separated from that date. Later related-story items on the page, such as the AlphaGo ten-year piece, are not used or relied on. Goldberg's running prediction is reported as his forecast, not as an outcome. The evidence kind is publisher-page, so no archive disclosure is needed.
“A robot that could outrun a human is probably right around the corner,” Goldberg says.
support — pass: Each claim is grounded in the source: the marketing role for startups and academics seeking lab funding, the need for disclosure to tell whether a robot is tele-operated, sideways squeezing of boxes, pristine settings, the outrun prediction, and carrying capacity with outstretched arms. The interpretive claim is labeled as Escape Meridian's reading and has a supporting excerpt. The limitations do not overstate the source.
for anyone hoping to ride the AI wave or demonstrate their progress—like a startup or an academic seeking lab funding—a good humanoid video is the best marketing tool available
media — pass: No media was selected, and no verified media candidates were supplied, so there is no media identity or relevance issue.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.