Amazon shuts down Mechanical Turk: the 'artificial artificial intelligence' AI finally killed, after 21 years

Amazon quietly closed the book on an era: Mechanical Turk, the 21-year-old crowdsourcing platform, will shut down for good on September 30, 2026. The backstage engine that helped train the entire machine-learning age finally fell at the feet of the AI it fed.

Three Key Facts

The end of an era. MTurk launched in 2005, letting businesses break work into "Human Intelligence Tasks" — data labeling, audio and video transcription, survey completion — and farm them out to workers worldwide, once serving over 500,000 of them. Amazon says the shutdown followed an internal assessment.

The end was telegraphed. The platform had been in decline for years, with Amazon investing few resources into it. It stopped accepting new customers on July 30; this is just the final cut.

A new labeling army takes over. Scale AI, Mercor, and Prolific — the new data-labeling upstarts — have already grabbed the work, recruiting the humans who feed frontier models. The irony writes itself: Bezos once called MTurk "artificial artificial intelligence," and now that real AI has arrived, its first move is to cut out the "artificial" part.

WangDou's Take

The dark comedy here is turned up to eleven: MTurk was the original "humans pretending to be AI" — hundreds of thousands of people hiding behind an API, faking being an algorithm, doing the dirty grunt work machines couldn't yet handle. Twenty years later, the machines learned the trick, then kicked those hundreds of thousands to the curb. This isn't an accident — it's the script. Every automation revolution burns people as fuel first, then tosses them once the fire is roaring. What's worth chewing on: AI didn't kill the human-labeling business, it just moved it from Amazon's hands into Scale's and Mercor's — nicer name, fatter valuation, same flesh-and-blood workers spoon-feeding the models backstage. The tech marches on; the pot at the bottom never changes, only the face carrying it gets younger.

Source: CNBC、TechCrunch