On Mechanical Turk

The new Slate/Economist podcast ‘The Secret History of the Future’ focuses on the lesser known stories of how some contemporary innovations are simply new versions of old ideas, which is a fine premis. This episode has been one of my favourites, discussing the idea of humans doing work that looks like it’s being performed by machines, which is a nicely unconventional theme. It begins with the story of the original ‘mechanical turk’, a clockwork chess-playing ‘machine’ invented in the 1880s that nearly beat one of the best chess players in the world.

Of-course, genius though the machine was, it wasn’t quite that genius. Inside the large wooden box that purportedly held all the clock-work machinery that could work the mechanical arm of a replica Turkish prince (hence ‘mechanical turk’) there was actually a (presumably non-clostrophobic) human who was making all the decisions.


The podcast discussion then moves on to focus on the modern day mechanical turk, the network of humans that perform various (often mundane) tasks for relatively low pay, and it doesn’t shy away from some of the moral issues that arise with these arrangements. But one idea really surprised me – the thought of services that we assume to be powered by a machine or an AI but which are actually underpinned by human input and labour. AIs being trained by the work of thousands of humans to classify and describe. ‘Chatbots’ that are actually humans. It had never occurred to me that this relatively hidden (but surely pretty new) phenomenon was a reality behind more services than we think.

Image By Carafe at English Wikipedia, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=7860391


电子邮件地址不会被公开。 必填项已用*标注