MIT Technology Review · Apr 20, 2026
MIT Technology Review reports Chinese tech workers train AI doubles and push back
Markdown · JSON · Original source
MIT Technology Review published this report on April 20, 2026. The publication describes Chinese tech workers being asked by bosses to document workflows so can automate tasks, alongside a backlash over what that process means for workers. A spoof project, Colleague Skill, went viral by claiming it could turn colleagues’ work histories into AI stand-ins. Worker Amber Li says using it to recreate a former coworker felt “uncanny and uncomfortable.” The article also reports that some workers see workflow documentation as reductive, while Emory University’s Hancheng Cao says firms can gain data about employee know-how and which work depends on human judgment. Koki Xu’s “anti-distillation” tool deliberately turns workflow material into generic language, pushing back against making a person’s skills reproducible. The report leaves a tension unresolved: Li says her company has not found a way to replace workers with AI tools, which she says remain unreliable and need constant supervision, yet she feels her value is being cheapened. How employees’ personality, tone, and judgment should be treated when companies create work agents remains unclear.
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: Apr 20, 2026 (2026-04-20T09:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-07T10:33:24.752Z. First observed: 2026-10-27T04:06:32.542Z. Retrieved: 2026-10-27T04:07:36.570Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 27, 2026; originally published April 20, 2026. The software engineer who describes training an AI on their workflow is unnamed; the article says they spoke anonymously because of concerns about job security. The article reports that ownership of personality, tone, and judgment is unclear; it does not resolve the legal questions it raises.
Open question
How should workers and companies determine who owns personality, tone, and judgment captured in AI work agents?
Source-stated
MIT Technology Review reports that bosses in China encourage tech workers to document workflows to automate tasks with AI agent tools.
their bosses are encouraging them to document their workflows in order to automate specific tasks and processes using AI agent tools like OpenClaw or Claude Code.
Source-stated
The article says Colleague Skill was created as a spoof, but its popularity prompted debate about workers’ dignity and individuality.
Though the project was created as a spoof, it struck a nerve among tech workers
Source-stated
Emory University assistant professor Hancheng Cao says companies can gain data on employee know-how and identify work that depends on human judgment.
Firms gain not only internal experience with the tools but also richer data on employee know-how, workflows, and decision patterns.
Source-stated
Li says her company has not found a way to replace workers with AI tools, which remain unreliable and require constant supervision; she feels her value is being cheapened.
largely because they remain unreliable and require constant supervision. “I don’t feel like my job is immediately at risk,” she says. “But I do feel that my value is being cheapened, and I don’t know what to do about it.”
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.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
Editorial provenance
Version 1. Publication ID: efccc47e-20ea-fd37-811a-160c12052c40.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-07T10:29:45.919Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_screen.
Observation: df4f3ba5840c06419878680b8f93b8d452aba6438eceb4d5880e1ee354b6a3c4. Source text hash: a5033e3ec63521f5eef672b56ec0d9516abb311a40ca721feecc52fdcf98cf7a.
Revision and source-check history
Version 1 · 2026-11-07T10:33:24.752Z
MIT Technology Review reports Chinese tech workers train AI doubles and push back
MIT Technology Review published this report on April 20, 2026. The publication describes Chinese tech workers being asked by bosses to document workflows so can automate tasks, alongside a backlash over what that process means for workers. A spoof project, Colleague Skill, went viral by claiming it could turn colleagues’ work histories into AI stand-ins. Worker Amber Li says using it to recreate a former coworker felt “uncanny and uncomfortable.” The article also reports that some workers see workflow documentation as reductive, while Emory University’s Hancheng Cao says firms can gain data about employee know-how and which work depends on human judgment. Koki Xu’s “anti-distillation” tool deliberately turns workflow material into generic language, pushing back against making a person’s skills reproducible. The report leaves a tension unresolved: Li says her company has not found a way to replace workers with AI tools, which she says remain unreliable and need constant supervision, yet she feels her value is being cheapened. How employees’ personality, tone, and judgment should be treated when companies create work agents remains unclear.
Original source
MIT Technology Review · Article
Source details
Original source date: Apr 20, 2026 (2026-04-20T09:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-07T10:33:24.752Z. First observed: 2026-10-27T04:06:32.542Z. Retrieved: 2026-10-27T04:07:36.570Z.
Limits
Text as retrieved on October 27, 2026; originally published April 20, 2026. The software engineer who describes training an AI on their workflow is unnamed; the article says they spoke anonymously because of concerns about job security. The article reports that ownership of personality, tone, and judgment is unclear; it does not resolve the legal questions it raises.
Open question
How should workers and companies determine who owns personality, tone, and judgment captured in AI work agents?
Source-stated
MIT Technology Review reports that bosses in China encourage tech workers to document workflows to automate tasks with AI agent tools.
their bosses are encouraging them to document their workflows in order to automate specific tasks and processes using AI agent tools like OpenClaw or Claude Code.
Source-stated
The article says Colleague Skill was created as a spoof, but its popularity prompted debate about workers’ dignity and individuality.
Though the project was created as a spoof, it struck a nerve among tech workers
Source-stated
Emory University assistant professor Hancheng Cao says companies can gain data on employee know-how and identify work that depends on human judgment.
Firms gain not only internal experience with the tools but also richer data on employee know-how, workflows, and decision patterns.
Source-stated
Li says her company has not found a way to replace workers with AI tools, which remain unreliable and require constant supervision; she feels her value is being cheapened.
largely because they remain unreliable and require constant supervision. “I don’t feel like my job is immediately at risk,” she says. “But I do feel that my value is being cheapened, and I don’t know what to do about it.”
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.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
Editorial provenance
Version 1. Publication ID: efccc47e-20ea-fd37-811a-160c12052c40.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-07T10:29:45.919Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_screen.
Observation: df4f3ba5840c06419878680b8f93b8d452aba6438eceb4d5880e1ee354b6a3c4. Source text hash: a5033e3ec63521f5eef672b56ec0d9516abb311a40ca721feecc52fdcf98cf7a.
passed · 2026-11-07T10:29:45.919Z · anthropic/claude-opus-5.5
Publication: efccc47e-20ea-fd37-811a-160c12052c40
attribution — pass: The reading attributes the report to MIT Technology Review and attributes each view to the right person: Li, Cao, Xu, Zhou's spoof project and the anonymous engineer. The closing remark that ownership is unclear reflects Xu's view in the article, and the limitations attribute it to the article. Li's 'uncanny and uncomfortable' is the article's indirect rendering of her words, which is acceptable.
making ownership much less clear
summary — pass: The reading leads with the development: bosses asking workers to document workflows for AI agents, and the backlash. It explains Colleague Skill as a viral spoof and describes the anti-distillation tool's function in plain language. 'AI agent' is defined on the page. It preserves the tension Li voices. It omits Li's finding that the tool was 'surprisingly good', but that does not distort the reading. The question follows from Xu's raised legal issue.
a skill like this can also capture elements of personality, tone, and judgment
date — pass: originalPublished is 2026-04-20T09:00:00Z, which matches the stated April 20, 2026 and the URL path. The page carries a dateModified of April 21, 2026, a day later, which does not change the original date. The reading uses no content from the trailing related-stories sidebar and has no hindsight framing. It is a publisher page, so no archive disclosure is needed.
support — pass: Each claim matches the source text. The source reports that bosses encourage workflow documentation and that Colleague Skill was a spoof that sparked a debate about dignity. Cao's remark about know-how and human judgment, Xu's sabotage tool that produces generic language, and Li's reliability caveat with her feeling of cheapened value are all in the text. No unsupported benefits, costs or forecasts are added.
the agent rewrites the material into generic, non-actionable language that would produce a less useful AI stand-in
media — pass: No media was selected, and no verified candidates were supplied. Leaving media empty is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
An AI system that chooses and carries out actions toward a goal, often using software tools.
Definition referenceAn AI system that chooses and carries out actions toward a goal, often using software tools.
Definition reference