Monday, February 27, 2017

Celebrating Natural Stupidity

I have been reading "The Undoing Project", Michael Lewis' gripping story of the collaboration between Daniel Khnemann and Amos Tversky, which revolutionized our understanding of decision-making and risk. In addition to the insights scattered throughout the book, the book also includes many of the witty aphorisms that Tversky was famous for, such as when he told the physicist Murray Gell-Mann, "Murray, no one is as smart as you think you are!"

One that I think is particularly profound (and close to home), is Tversky saying: "My colleagues, they study artificial intelligence; me, I study natural stupidity."

Not only is it funny, it is also something that those of us working in AI need to pay attention to. Most of us take it for granted that the goal of AI is to build machines that can reason better, integrate more data, make more rational decisions. What the work of Kahnemann and Tversky shows us is that this is not how people (or animals) function. If the goal in artificial intelligence is to replicate human capabilities, we should also be considering the natural stupidity of our machines. Of course, one also has to think of what Kahnemann calls "thinking fast and slow", that is, the existence of an intuitive "fast thinking" system, and a more rational "slow thinking" system. There is much more to be said about this, and I hope to write about that at greater length soon....

I would recommend the book to anyone interested in anything interesting

No comments:

Post a Comment