Amazon’s Mechanical Turk: Artificial Artificial Intelligence
Clive Thomson has an interesting article on Mechanical Turk, a new Amazon company. Mechanical Turk provides developers with an interface for solving problems that are better solved by humans. As Clive puts it:
Computers suck at many tasks that are super-easy for humans. Any idiot can look at picture and instantly recognize that it’s a picture of a pink shoe. Any idiot can listen to a .wav file and realize it’s the sound of a dog barking. But computer scientists have spent billions trying to train software to do this, and they’ve utterly failed.
Mechanical Turk provides an interface for people to complete simple tasks like these, and get paid for their work (albeit very little). An example of a successful business that uses Mechanical Turk is CastingWords, a podcast transcription service. Computerized voice recognition is still shaky, especially for something like a podcast, whereas humans can do it quite easily. Other examples of tasks in which Mechanical Turk can assist are image recognition, surveys, or aesthetic evaluations.
Each task usually pays anywhere from a few cents to a few dollars, so it’s unlikely that the humans completing these tasks could make much of a living at it. But a few minutes here and there could bring in a few extra dollars a day. Interestingly, the article also mentions potential labor issues, as Mechanical Turk provides a “virtual day laborer hiring hall” for things that companies might otherwise pay wages for.
I’d be curious to here feedback from users, either from the developer or the worker side.














