Podcast: Deception In AI Development And The Role Of Humans

Around 1770, a Hungarian inventor Wolfgang von Kempelen constructed a chess-playing machine called the Mechanical Turk, which consisted of a life-sized model of a human dressed in Turkish robes sitting at a chess table. The device won most of the games played during its demonstrations around the world, defeating such opponents as Napoleon Bonaparte or Benjamin Franklin. Many contemporaneous engineers tried to figure out the machine, but each failed to pin down how the automation worked. What they didn’t know was that the Turk was merely an elaborate deception: it was a human chess master concealed inside the construction that was responsible for all the winning moves. And although an actual machine – IBM’s Deep Blue – managed to win a chess game against a reigning world champion Garry Kasparov in the famous match of May over two hundred years later, there are still plenty of areas where human input is indispensable to perform tasks that computers are currently unable to do.

By drawing insightful parallels to the Kempelen’s invention, the Economist’s Tom Standage and Slate’s Seth Stevenson explore the role of humans and deception in the development of artificial intelligence. As they do it, they examine data-practices of the modern day tech companies, covering such applications as Mechanical Turk, a market for human intelligence created by Amazon, or Captcha systems that help train AIs to get smarter. The result is an absorbing 37-minute podcast, the first of a new series The Secret History of the Future that aims to “examine the history of tech to uncover stories that help us illuminate the present and predict the future.” And that's exactly what Slate and the Economist's collaboration debut does. “The story of the 18th-century chess machine turns out to be one of those curious tales from history that can help us understand technology today, and where it might go tomorrow,” the episode’s introduction says.