Platform microtask laborers train AI that will eventually displace them

Evidence

The Turker is perhaps the quintessential example. Turkers are freelance independent contractors who perform small tasks on Amazon’s Mechanical Turk platform, or MTurk, a crowd sourcing marketplace launched in 2005 where Requesters (businesses or individuals) can post tasks that computers can’t currently perform or that are cheaper for humans to do. Armies of Turkers can then earn small amounts of money (usually less than one cent per task) to perform the jobs, which might include doing things such as categorizing, sorting, testing, rating, generating content, or training artificial intelligence algorithms that will eventually make the human Turker redundant. Workers are seemingly empowered because, as contractors, they set their own hours and choose their own tasks. But employers are empowered too, since they don’t have to pay payroll taxes, can avoid regulations concerning minimum wage and overtime work, and can get millions of microtasks performed that regular employees might resist doing. — Nick Couldry & Ulises A. Mejias, The Costs of Connection, 217

Categorization

• QUESTION: The workers-training-their-own-replacement loop is a compressed parable worth unpacking: does the gig/crowd model structurally depend on labor that erases its own future demand?
• SEED: Possible essay angle: ‘the surplus labor paradox’ — gig workers subsidize the automation that terminates their own gigs — as a through-line connecting Amazon’s two faces (warehouse and Turk).
• XREF: connects to: Algorithmic services extract unpaid user labor that trains the very algorithms — Yanis Varoufakis, Technofeudalism: What Killed Capitalism, 315; Algorithms on factory floors merely continue Taylorist machine pacing of workers — Yanis Varoufakis, Technofeudalism: What Killed Capitalism, 323; Microtask workers train the AI systems that will replace them — Nick Couldry & Ulises A. Mejias, The Costs of Connection: How Data Is…, 217 (+6 more)

Broader Connections

• Microtask workers train the AI systems that will replace them #corroborates (1.00) — Nick Couldry & Ulises A. Mejias, The Costs of Connection, 217
• AI reduces job quality by devaluing human labor. #expands (0.75) — Emily M. Bender, Alex Hanna, The AI Con- How to Fight Big Tech’s…, 126
• Generative AI threatened to devalue the very concept of labor #expands (0.72)
• AI threatens to devalue labor itself, not just displace workers #expands (0.72) — W. David Marx, Blank Space, 693
• Platform users become unpaid servants who train algorithms. #exemplifies (0.70)
• AI’s employment consequences are unpredictable because they depend on capitalist competition #complicates (0.70) — Clara E. Mattei, Escape from Capitalism, 248
• Delivery algorithms capture workers’ tacit route knowledge and render them replaceable. #exemplifies (0.70) — Donatella della Porta et al., Mobilizing Against the Odds, 102