Articles and Essays
A robot that doesn't recognize a Black face. A self-driving car that's better at recognizing white predestrians than people of color. They're products of institutions where Black researchers get asked what they're doing in the Comp Sci building and do they need directions. I spoke with roboticists and computer-science researchers are debating what to do about systemic racism.
Note to Google Chrome users (10-12-20): For reasons unknown the links to these writings will not appear in your browser. To see links please load the page in a different browser.
Why the pandemic has been good for robotics
My cover story for the September, 2020 issue of National Geographic.
Getting humans and machines to work well together isn't easy
When a "deep learning" algorithm makes a mistake, we can't tell why. My 2015 piece on adversarial examples and algorithmic opacity
You clicked "Agree" but you didn't really read the contract, did you? Nobody does, and that matters.
It's possible to delegate too much to robots and algorithms
Studies show people value secret information more just because it is secret. No wonder we have a problem with "overclassification."
All good people agree,
And all good people say,
All nice people, like Us, are We
And every one else is They.
-Rudyard Kipling, "A Friend of the Family"
Scientists, when they turn their attention to people, usually talk about the entire human race or about the individual human being. Those are two faces of the same idea. Truth about all is truth about each . . .
A dystopian tale about the uses of medical genetics.
A FAQ about the book.
Machines are getting smarter all the time. Animals are turning out to have unexpected mental powers, too. So what happens to our sense of the dividing line between human and non-human?
By definition, studying human kinds means bridging disciplinary boundaries. Human kinds involve bodily states, like stress levels, and brain activity, and mental states, and the political and cultural world that each mind shares with others. This article describes one such cross-cultural project: Collaboration among neuroscientists, brain scanners and social psychologists to look at how a brain function can be tied to human-kind behavior. Specificially, these two studies looked at differences in brain activity when people looked at members of different races.
Language is the most emotionally powerful of our tools for deciding who is Us and who is Them, and which human kinds are real. So fights over language politics are fierce. This piece provoked more angry email than almost anything else I've ever written. It also provoked some intelligent reflections, though, one of which made me change my mind about a few things.
A first-hand lesson in how we don't get to choose where we stand in a tribal fight.
A built-in predisposition to see human kinds, argues Lawrence Hirschfeld, makes it easy for little kids to learn racial categories.
For John James Audubon, the great painter of wildlife and champion of American nature, a good day was one in which he shot a hundred birds. That was how he got the relentless see-it-all detail in those huge prints that made his fortune.
For many biologists, as for him, loving nature means wanting to understand it, and understanding it requires control over a bit of it. That often entails what the journal papers will still refer to as a ``sacrifice'' of laboratory creatures. The word's not lightly chosen. Many biologists admire and cherish the animals they kill. No doubt farmers do too (or did, when farms were not factories) but in science the compensating value for killing animals is more abstract. The ``sacrifice'' produces knowledge, not food. Knowledge the scientist wants because, at root, she loves life and admires animals.
During the time when I was thinking about this, I coincidentally had a chance to experience it first-hand. That's what this essay is about.
Do physicists ``invent'' things like atoms and quarks, or do they ``discover'' them? What's the link between equations in a journal and reality? On such issues physicists themselves are all over the map.
You don't have to look far to find physicists who say their work depicts an ultimate reality. Google Steven Weinberg or Frank Wilczek, and you'll see. On the other hand, it is not hard to find physicists who believe the opposite -- that we can't say there is a final truth out there without reference to the minds that seek it. Try Peter Galison or Piet Hut on that Google search, and you'll see what I mean.
Some years ago I had the privilege of encountering one of physics' great minds, who addressed this subject (and countless others) in our interviews. This ran in the New York Times Magazine.
As activists try to create human kinds that suit them, they often meet resistance. Arguments over Attention Deficit Disorder are a good illustration of ongoing contention over how this human kind should be defined -- and who should define it.
What's the difference between a theory and a fact?
I wrote a few articles in the 1990's about human kinds that were being created, or redefined, by focussed, determined work. How was it possible that people could become convinced they belonged to a tribe they never heard of, or thought of as a medical condition, or even a moral flaw? (I now think the answer is the mind's innate mechanism for dealing with human kinds.)
This is one of those articles -- about the movement to make dwarfism less of a medical diagnosis and more of a tribal identity.
Why American Hindus Were ``White'' in 1910, but not in 1920. The wide range of racial and ethnic definitions in census information reveals how malleable are the definitions of race and ethnicity, from time to time, and place to place. That was the theme of this review-essay about Clara Rodriguez' fascinating book on the subject.
Some thinkers try to separate human kinds into those with a physical basis and those that are ``merely'' cultural. Being a Christian Democrat is cultural, in this view, while being a woman is biology.
I don't think this distinction makes much sense. Even biological categories have culture in their roots, for one thing. And cultural categories often have some biological traits wrapped into their folds. This is true even of the most biological of all the obvious human-kind differences. That's what this article describes.
Human-kind feelings are the link between body, mind and society. Jonathan Shay is a psychiatrist who has studied how these connections can go wrong for soldiers in combat -- and what has to be done to prevent it.
How Colleges Become Tribes, or, What Football Rallies, Hazing, Secret Societies, School Songs and so on reveal about the mind
One of the most familiar examples of joining a new tribe is the prestigious American college, where new students quickly learn that students and alumni of their school are a breed apart. It happens within the school, too, as students sort themselves into fraternities, sororities and other human kinds. This is very handy for their professors, who sometimes use school spirit and its symbols -- pennants, songs, bonfires -- as examples of human-kind behavior.
What do we learn from experience, and what are we born with? The notion that we’re all ``blank slates’’ has taken quite a beating in the past 50 years, thanks to evidence that the mind comes pre-set to create certain kinds of perceptions, and to figure out very complicated patterns. Linguistics is where this ``innatist’’ work began, and this article for Discover was my first encounter with these ideas. And with Derek Bickerton, who proposed that the best way to see how children acquire a new language was to put them all on a desert island and see . . .
Well, you can see for yourself.