As I said before many times, and as I
will say many times in the future, we have to design systems for people the
way they are not, the way we want them to be. We want to design technology to be a
collaborator, a team worker, with people. And yet, we still think that people are somehow deficient,
and we have to replace them with machines. Look, recently Daniel Kahneman, Danny,
wrote a book called Thinking, Fast and Slow, and has gotten a tremendous amount of press. The
work of Kahneman and his collaborator, Amos Tversky, did a lot of work showing the biases
that people have. Well, I disagree with my good friend Danny:
we don't have biases. This is the way we work. This is the way we're genetically
predisposed. It has actually been very, very valuable in the evolution and
success of human beings. For example, we have a recency bias -- we tend to be
biased by what happened to us most recently. Well, that makes good sense, actually.
The most recent events are the ones most likely to be repeated or to
impact what is happening. We also have other kinds of biases, and they're all
well-described in this book. But it's kind of an artificial book, because the
book is all about situations that you're asked to make judgments on
out of context. Put us in context, and these biases are often positives. Recently, a designer friend wrote that, "Oh,
this is an exciting book, Thinking, Fast and Slow," and actually, it is, it's a fine book. But, he said, "Why don't we design our
technology that will correct the biases of people? So, we'll instruct people and we'll help
people to overcome their biases." I say no; because the biases are normal
ways that people have operated. Most of the time, they actually are beneficial.
It's only when some clever experimental psychologists does a study says, "See! See how silly these people were? They made their own judgment." Ack! They made their own judgment in a problem that
would almost never, ever occur in real life. So... what I say: why not do it the other way around?
Why not take advantage of the way that people work? Let's make technology that actually supports us. Where
those judgments of people are sometimes wrong, because they're biased too heavily,
by say a recent event or by unique events, Well, then let the system say, "Here's a suggestion," or "Have you considered this other possibility." Let the system work with us, but don't call them biases
and erroneous judgments and bad decision-making. Take it for human judgments and human decision-making, which most of the time, is extremely powerful, and most of the time, is correct.
