For about 85 years the strange automaton called the Mechanical Turk toured Europe and America, drawing crowds. It appeared to play chess on its own, but a hidden human inside revealed the secret. Charles Michael Carroll explains that the era of the industrial revolution made mechanical marvels a daily miracle, so people were eager for mystification and charlatanism.

In the 21st century the name resurfaced as a crowd‑labour platform. Amazon Mechanical Turk fed small, low‑paid tasks to freelancers. An American warehouse worker earned $5 to $7 an hour transcribing audio in his “man cave”. These tasks also fed data to train AI systems.

As large language models grew capable, the need for such human workers declined. Amazon announced that Amazon Mechanical Turk will close to new customers on July 30 2026. A worker on Reddit wrote, “Training those AIs, looks like we digged our own grave.”

The demand is shifting to physical‑world training—factory workers in India with cameras on their heads—and to niche expertise. AI‑training firms like Mercor pay $80 to $120 an hour for “journalism evaluators” and “finance evaluators” who grade AI output. Crowd platforms now focus on tasks where human data is still essential, while workers use AI to speed up their own work. Synthetic sampling, where LLMs role‑play as humans, is emerging, offering faster, cheaper data but raising questions.