The AI labelled the image. A human in the Philippines labelled the image. The AI got the credit.
Literal meaning Mechanical Turk — a famous 18th-century chess automaton that appeared to play autonomously but concealed a human chess master inside. Amazon’s Mechanical Turk (MTurk) is a crowdsourcing platform that names itself after this deception.
Origin Amazon launched MTurk in 2005, offering businesses access to a global on-demand workforce for “Human Intelligence Tasks” — microtasks that algorithms cannot yet perform: image labelling, content moderation, data verification. Hara et al. (2018) documented that median wages on the platform are well below minimum wage in any jurisdiction. The name — after a device designed to make human labour invisible — is analytically exact.
Amazon’s crowdsourcing platform for human microlabour — named after a historical deception it exactly reproduces: human intelligence presented as machine intelligence.
The Appeal MTurk provides genuine income and flexible work for participants globally, and enables AI development that would otherwise be impossible or prohibitively expensive.
The Friction Hara et al. (2018) document earnings averaging $2/hour with no labour protections, sick pay, or appeal mechanisms. Shapiro (2024) frames MTurk as a “digital sweatshop” that stripped workers of all protections except those they built themselves. Ghost Work is the broader category: invisible human labour that keeps AI systems functional while being structurally erased from the product narrative.
Why This Matters Once you see Mechanical Turk as a labour infrastructure rather than a technology platform, you see that the AI systems presented as autonomous are dependent on an invisible, poorly paid, unprotected global workforce.
Related terms: Ghost Work · Surveillance Capitalism · Technofeudalism · Deskilling
Read more:
- A Data-Driven Analysis of Workers’ Earnings on Amazon Mechanical Turk — Hara, K. et al. (2018). CHI Conference on Human Factors in Computing Systems
- Amazon Mechanical Turk: The Human Sciences’ Labor Problem — Shapiro, A. (2024). Labor: Studies in Working-Class History