{"id":"7231f316f469aa34b3a8a94677a39ee9e9fe8723649217022835f2c37b441ba3","title":"MIT Technology Review reports on collecting movement data to train humanoid robots","argument":"MIT Technology Review published this account on April 21, 2026. The publication describes companies gathering recordings and sensor data of people doing physical tasks, betting that movement data can help train humanoid robots. The collection ranges from people filming chores at home to workers repeating tasks in training centers; one US delivery company has also equipped employees with movement-tracking sensors, partly to study injuries and partly to train robots that could replace them.\n\nThe publication connects this effort to the challenge of adapting training approaches used for large language models to robotics: companies lack an internet-scale collection of data about human movement, and simulations do not perfectly reproduce real-world friction or elasticity. Its central tension is that physical laborers may increasingly become data collectors, while it remains unclear whether movement data can be gathered at the scale needed for technical breakthroughs or support a profitable business. The account ends by asking how many ordinary task recordings it might take to teach a robot to cook dinner, and suggests this may be the year that question gets an answer.","topicIds":["work-and-business--jobs-and-occupations","job-displacement-and-creation"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on November 8, 2026; originally published April 21, 2026. The retrieved text does not identify the account’s author by name.","speaker":{"name":"MIT Technology Review","roleAtSource":"Publisher"},"source":{"title":"Humanoid data","publisher":"MIT Technology Review","format":"Article","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","publishedAt":"2026-04-21T20:45:00Z","dateBasis":"Date from the publisher’s own records, through its API."},"evidence":[{"text":"MIT Technology Review says robotics companies are collecting data about how people move to train humanoid robots.","excerpt":"robotics companies are betting that data about the way we move will help them build more capable humanoid robots","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication says simulations do not perfectly model real-world friction or elasticity, and robots trained in them tended to stumble.","excerpt":"simulations never perfectly model how things like friction or elasticity work in the real world, so the robots trained in them tended to (literally) stumble.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication reports that gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","excerpt":"Gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication says a US delivery company equipped employees with sensors to track their movements, partly to study injuries and also with the goal of training robots that could replace them.","excerpt":"a delivery company in the US had outfitted its employees with sensors that track their movements as they carry boxes, in part to study injuries but also with the goal of training robots that could replace them.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"MIT Technology Review says it is unclear whether movement data can be gathered at the scale potentially needed for technical breakthroughs or support a profitable business.","excerpt":"It’s not clear that it’s even possible to do it at the scale potentially needed to yield technical breakthroughs, let alone build a profitable business.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication reports that humanoid robotics alone drew $6.1 billion in venture capital in 2025.","excerpt":"$6.1 billion in 2025 for humanoids alone","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How much movement data would it take to teach a robot to perform a complex household task such as cooking dinner?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MIT Technology Review reports on collecting movement data to train humanoid robots","summary":"MIT Technology Review published this account on April 21, 2026. The publication describes companies gathering recordings and sensor data of people doing physical tasks, betting that movement data can help train humanoid robots. The collection ranges from people filming chores at home to workers repeating tasks in training centers; one US delivery company has also equipped employees with movement-tracking sensors, partly to study injuries and partly to train robots that could replace them.\n\nThe publication connects this effort to the challenge of adapting training approaches used for large language models to robotics: companies lack an internet-scale collection of data about human movement, and simulations do not perfectly reproduce real-world friction or elasticity. Its central tension is that physical laborers may increasingly become data collectors, while it remains unclear whether movement data can be gathered at the scale needed for technical breakthroughs or support a profitable business. The account ends by asking how many ordinary task recordings it might take to teach a robot to cook dinner, and suggests this may be the year that question gets an answer.","question":"How much movement data would it take to teach a robot to perform a complex household task such as cooking dinner?","claims":[{"text":"MIT Technology Review says robotics companies are collecting data about how people move to train humanoid robots.","excerpt":"robotics companies are betting that data about the way we move will help them build more capable humanoid robots","support":"source-stated"},{"text":"The publication says simulations do not perfectly model real-world friction or elasticity, and robots trained in them tended to stumble.","excerpt":"simulations never perfectly model how things like friction or elasticity work in the real world, so the robots trained in them tended to (literally) stumble.","support":"source-stated"},{"text":"The publication reports that gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","excerpt":"Gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","support":"source-stated"},{"text":"The publication says a US delivery company equipped employees with sensors to track their movements, partly to study injuries and also with the goal of training robots that could replace them.","excerpt":"a delivery company in the US had outfitted its employees with sensors that track their movements as they carry boxes, in part to study injuries but also with the goal of training robots that could replace them.","support":"source-stated"},{"text":"MIT Technology Review says it is unclear whether movement data can be gathered at the scale potentially needed for technical breakthroughs or support a profitable business.","excerpt":"It’s not clear that it’s even possible to do it at the scale potentially needed to yield technical breakthroughs, let alone build a profitable business.","support":"source-stated"},{"text":"The publication reports that humanoid robotics alone drew $6.1 billion in venture capital in 2025.","excerpt":"$6.1 billion in 2025 for humanoids alone","support":"source-stated"}],"limitations":["Text as retrieved on November 8, 2026; originally published April 21, 2026.","The retrieved text does not identify the account’s author by name."],"mediaId":null,"correctionReason":null,"id":"7231f316f469aa34b3a8a94677a39ee9e9fe8723649217022835f2c37b441ba3","publicationId":"e646e2aa-1fd4-d1e4-6391-9ed811254a70","version":1,"sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","sourceTitle":"Humanoid data","publisher":"MIT Technology Review","publishedAt":"2026-04-21T20:45:00Z","originalPublished":{"value":"2026-04-21T20:45:00Z","precision":"instant","basis":"backfill:api_published"},"firstObservedAt":"2026-11-08T21:27:51.841Z","createdAt":"2026-11-09T00:38:28.181Z","historical":true,"entityIds":[],"topicIds":["work-and-business--jobs-and-occupations","job-displacement-and-creation"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"6fdea3be7a7cdab80c4afbe022e6a7a1f656a83836d8106de99dfca8bf3f1755","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.71,"decidedAt":"2026-11-08T21:27:51.377Z"},"provenance":{"textHash":"51c50704a712082544c6e4c16c52ca148c3b496e37e89889106f5029a9b29fe5","observationId":"3ce5b6a4b8db2ffda35e80a83d7f01c8f43dab99046b76e9a06360f8db81fbc9","retrievedAt":"2026-11-08T21:29:36.266Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"25be758e-9422-46ff-851d-08ea9447046a"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"dfd10f9c-1b89-f8ab-f919-decd8e3f226f"}},"audit":{"status":"passed","reviewedAt":"2026-11-09T00:34:19.622Z","model":"anthropic/claude-opus-5.5"}},"canonicalUrl":"https://escapemeridian.com/readings/mit-technology-review-reports-on-collecting-movement-data-to-7231f316","formats":{"html":"/readings/mit-technology-review-reports-on-collecting-movement-data-to-7231f316","markdown":"/readings/mit-technology-review-reports-on-collecting-movement-data-to-7231f316.md","json":"/readings/mit-technology-review-reports-on-collecting-movement-data-to-7231f316.json"},"attribution":"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.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-llm","name":"Large language model","explanation":"A language model with many learned numerical settings, used to process and generate language.","referenceUrl":"https://developers.google.com/machine-learning/glossary#large-language-model"}],"entities":[]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"7231f316f469aa34b3a8a94677a39ee9e9fe8723649217022835f2c37b441ba3","revision":"e646e2aa-1fd4-d1e4-6391-9ed811254a70","revisionKind":"publication-id","title":"MIT Technology Review reports on collecting movement data to train humanoid robots","url":"https://escapemeridian.com/readings/mit-technology-review-reports-on-collecting-movement-data-to-7231f316"},"source":{"type":"source","id":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","revision":"51c50704a712082544c6e4c16c52ca148c3b496e37e89889106f5029a9b29fe5","revisionKind":"sha256-source-text","title":"Humanoid data","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"7231f316f469aa34b3a8a94677a39ee9e9fe8723649217022835f2c37b441ba3","title":"MIT Technology Review reports on collecting movement data to train humanoid robots","argument":"MIT Technology Review published this account on April 21, 2026. The publication describes companies gathering recordings and sensor data of people doing physical tasks, betting that movement data can help train humanoid robots. The collection ranges from people filming chores at home to workers repeating tasks in training centers; one US delivery company has also equipped employees with movement-tracking sensors, partly to study injuries and partly to train robots that could replace them.\n\nThe publication connects this effort to the challenge of adapting training approaches used for large language models to robotics: companies lack an internet-scale collection of data about human movement, and simulations do not perfectly reproduce real-world friction or elasticity. Its central tension is that physical laborers may increasingly become data collectors, while it remains unclear whether movement data can be gathered at the scale needed for technical breakthroughs or support a profitable business. The account ends by asking how many ordinary task recordings it might take to teach a robot to cook dinner, and suggests this may be the year that question gets an answer.","topicIds":["work-and-business--jobs-and-occupations","job-displacement-and-creation"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on November 8, 2026; originally published April 21, 2026. The retrieved text does not identify the account’s author by name.","speaker":{"name":"MIT Technology Review","roleAtSource":"Publisher"},"source":{"title":"Humanoid data","publisher":"MIT Technology Review","format":"Article","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","publishedAt":"2026-04-21T20:45:00Z","dateBasis":"Date from the publisher’s own records, through its API."},"evidence":[{"text":"MIT Technology Review says robotics companies are collecting data about how people move to train humanoid robots.","excerpt":"robotics companies are betting that data about the way we move will help them build more capable humanoid robots","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication says simulations do not perfectly model real-world friction or elasticity, and robots trained in them tended to stumble.","excerpt":"simulations never perfectly model how things like friction or elasticity work in the real world, so the robots trained in them tended to (literally) stumble.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication reports that gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","excerpt":"Gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication says a US delivery company equipped employees with sensors to track their movements, partly to study injuries and also with the goal of training robots that could replace them.","excerpt":"a delivery company in the US had outfitted its employees with sensors that track their movements as they carry boxes, in part to study injuries but also with the goal of training robots that could replace them.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"MIT Technology Review says it is unclear whether movement data can be gathered at the scale potentially needed for technical breakthroughs or support a profitable business.","excerpt":"It’s not clear that it’s even possible to do it at the scale potentially needed to yield technical breakthroughs, let alone build a profitable business.","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"},{"text":"The publication reports that humanoid robotics alone drew $6.1 billion in venture capital in 2025.","excerpt":"$6.1 billion in 2025 for humanoids alone","url":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How much movement data would it take to teach a robot to perform a complex household task such as cooking dinner?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MIT Technology Review reports on collecting movement data to train humanoid robots","summary":"MIT Technology Review published this account on April 21, 2026. The publication describes companies gathering recordings and sensor data of people doing physical tasks, betting that movement data can help train humanoid robots. The collection ranges from people filming chores at home to workers repeating tasks in training centers; one US delivery company has also equipped employees with movement-tracking sensors, partly to study injuries and partly to train robots that could replace them.\n\nThe publication connects this effort to the challenge of adapting training approaches used for large language models to robotics: companies lack an internet-scale collection of data about human movement, and simulations do not perfectly reproduce real-world friction or elasticity. Its central tension is that physical laborers may increasingly become data collectors, while it remains unclear whether movement data can be gathered at the scale needed for technical breakthroughs or support a profitable business. The account ends by asking how many ordinary task recordings it might take to teach a robot to cook dinner, and suggests this may be the year that question gets an answer.","question":"How much movement data would it take to teach a robot to perform a complex household task such as cooking dinner?","claims":[{"text":"MIT Technology Review says robotics companies are collecting data about how people move to train humanoid robots.","excerpt":"robotics companies are betting that data about the way we move will help them build more capable humanoid robots","support":"source-stated"},{"text":"The publication says simulations do not perfectly model real-world friction or elasticity, and robots trained in them tended to stumble.","excerpt":"simulations never perfectly model how things like friction or elasticity work in the real world, so the robots trained in them tended to (literally) stumble.","support":"source-stated"},{"text":"The publication reports that gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","excerpt":"Gig workers in Nigeria, Argentina, and India are filming themselves doing chores at home.","support":"source-stated"},{"text":"The publication says a US delivery company equipped employees with sensors to track their movements, partly to study injuries and also with the goal of training robots that could replace them.","excerpt":"a delivery company in the US had outfitted its employees with sensors that track their movements as they carry boxes, in part to study injuries but also with the goal of training robots that could replace them.","support":"source-stated"},{"text":"MIT Technology Review says it is unclear whether movement data can be gathered at the scale potentially needed for technical breakthroughs or support a profitable business.","excerpt":"It’s not clear that it’s even possible to do it at the scale potentially needed to yield technical breakthroughs, let alone build a profitable business.","support":"source-stated"},{"text":"The publication reports that humanoid robotics alone drew $6.1 billion in venture capital in 2025.","excerpt":"$6.1 billion in 2025 for humanoids alone","support":"source-stated"}],"limitations":["Text as retrieved on November 8, 2026; originally published April 21, 2026.","The retrieved text does not identify the account’s author by name."],"mediaId":null,"correctionReason":null,"id":"7231f316f469aa34b3a8a94677a39ee9e9fe8723649217022835f2c37b441ba3","publicationId":"e646e2aa-1fd4-d1e4-6391-9ed811254a70","version":1,"sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","sourceTitle":"Humanoid data","publisher":"MIT Technology Review","publishedAt":"2026-04-21T20:45:00Z","originalPublished":{"value":"2026-04-21T20:45:00Z","precision":"instant","basis":"backfill:api_published"},"firstObservedAt":"2026-11-08T21:27:51.841Z","createdAt":"2026-11-09T00:38:28.181Z","historical":true,"entityIds":[],"topicIds":["work-and-business--jobs-and-occupations","job-displacement-and-creation"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2026/04/21/1135656/humanoid-data-robot-training-ai-artificial-intelligence/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"6fdea3be7a7cdab80c4afbe022e6a7a1f656a83836d8106de99dfca8bf3f1755","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.71,"decidedAt":"2026-11-08T21:27:51.377Z"},"provenance":{"textHash":"51c50704a712082544c6e4c16c52ca148c3b496e37e89889106f5029a9b29fe5","observationId":"3ce5b6a4b8db2ffda35e80a83d7f01c8f43dab99046b76e9a06360f8db81fbc9","retrievedAt":"2026-11-08T21:29:36.266Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"25be758e-9422-46ff-851d-08ea9447046a"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"dfd10f9c-1b89-f8ab-f919-decd8e3f226f"}},"audit":{"status":"passed","reviewedAt":"2026-11-09T00:34:19.622Z","model":"anthropic/claude-opus-5.5"}}}],"audits":[{"publicationId":"e646e2aa-1fd4-d1e4-6391-9ed811254a70","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the account to MIT Technology Review, which matches the evidence. It notes that the retrieved text names no author, even though the piece is written in the first person. The claims are framed as the publication's statements, not as established fact.","excerpt":"I was recently invited to join an app that would pay me cryptocurrency to film myself doing tasks"},{"dimension":"summary","status":"pass","explanation":"The summary faithfully conveys the argument. Companies are betting that human movement data can train humanoids. Simulations fall short. Data collection ranges from gig workers to training centers to delivery-company sensors. The tension is laborers becoming data collectors, with scale and profitability unclear. The closing question and the hedged 'this year' line are preserved. Large language model is defined beside the reading, and other terms are plain. The question follows directly from the source.","excerpt":"All this points to a surreal future of work in which physical laborers increasingly become data collectors."},{"dimension":"date","status":"pass","explanation":"originalPublished is 2026-04-21T20:45Z, and April 21, 2026 matches its UTC calendar day. The 'this year' remark is reported as the source's suggestion, not as an outcome. The reading does not use the later sidebar list of related stories. The evidence kind is a publisher page, so no archive disclosure is needed.","excerpt":"Perhaps this’ll be the year we find out."},{"dimension":"support","status":"pass","explanation":"Each claim matches a verbatim passage: the movement-data bet, simulation limits, gig workers filming chores, delivery-company sensors, the uncertainty about scale and profit, and the $6.1 billion in 2025 for humanoids. No unsupported benefits, costs or forecasts are added.","excerpt":"as venture capital money poured into robotics—$6.1 billion in 2025 for humanoids alone"},{"dimension":"media","status":"pass","explanation":"No media is selected, and no verified candidates were supplied, so there is no media identity or relevance issue.","excerpt":null}],"createdAt":"2026-11-09T00:34:19.622Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}