I've seen the decline of Mechanical Turk linked to workers using AI and to AWS's position in AI. How much do those connections actually explain?
What does Mechanical Turk's decline have to do with AI?
Answers
AI use creates a pretty obvious problem for tasks where the buyer specifically wants human judgment. If a worker submits model generated text, the result may look fine but no longer be the kind of response the buyer paid for.
A 2023 study estimated that 33% to 46% of workers used large language models on a text production task. That's a striking result, but it's one task in one study, not a measurement of every task on the platform.
Reply to this answer
I wouldn't jump from that to saying AI caused the decline, or that it proves AWS is behind in AI. The worker study points to a challenge for one kind of task. It doesn't establish why the service has faded or how AWS compares with competitors.
Reply to this answer
There are possible ways to make unauthorized AI use harder, like monitored apps or locked down devices. But those are tradeoffs, not magic detection. They can add friction and privacy concerns, and they don't necessarily prove someone hasn't used another device.
Reply to this answer
For some customers, an AI option could be a practical replacement if it's clearly labeled and meets their needs. That wouldn't be the same product as human responses, though. Buyers would need to choose based on what the task actually requires.
Reply to this answer
Add your answer
Answers are reviewed before publication.
j537~ptr_8:
That distinction matters for research too. A study looking for human responses could get misleading results even if the generated text seems plausible.