4 ms·

"If some guy rolls in with a fancy algorithm telling you someone is a criminal using facial features, you need to be able to say that is nuts." OK, maybe. But

"If some guy rolls in with a fancy algorithm telling you someone is a criminal using facial features, you need to be able to say that is nuts." OK, maybe. But I think that it is just as often political correctness as science that allows one to dismiss such assertions out of hand. As much, as a gentile, I would like to be able to dismiss "The Jews really are smarter than the rest of us" out of hand, some unpopular assertions may in fact be true. And I think truth has to come before other considerations. (Though it may be best not to make such assertions at all if there's not a good reason to.) Overall, though, I really liked what Bergstrom had to say. What he had to say about mockery being unproductive was some practical wisdom.

a1369209993 6y ago
It's not that they are, in fact, not a criminal. That's probably true, since unless you're using a deeply fucked up juridiction's definition of criminal, most people aren't criminals, but that's not the point. What's nuts is treating the claims of a algorithm you don't understand as evidence that someone is a criminal, rather than as unsubstantiated hearsay the way you would if a human made such a claim.
- pmdulaney 6y ago
  If Bergstrom's point is that one shouldn't believe claims just because someone asserts them, then yes, that's true. But I would say that the set of assertions for which one is justified in rejecting them out of hand is perhaps smaller than he is suggesting. For example, there was fake news recently in which the Gates Foundation was said to be inserting subcutaneous chips into folks who had received vaccines. It turned out to be untrue, but not obviously so. The technology exists and if not abused it could be an inexpensive and effective form of record keeping.
  - a1369209993 6y ago
    > rejecting them out of hand I may have missed it, but I don't see TFA actually arguing for that at all. > one shouldn't believe claims just because someone [or something] asserts them Seems to be the overall point. If some guy rolls in with a paranoid schizophrenic telling you someone is a criminal using facial features, you need to be able to say that is nuts.

a1369209993 6y ago

It's not that they are, in fact, not a criminal. That's probably true, since unless you're using a deeply fucked up juridiction's definition of criminal, most people aren't criminals, but that's not the point. What's nuts is treating the claims of a algorithm you don't understand as evidence that someone is a criminal, rather than as unsubstantiated hearsay the way you would if a human made such a claim.

pmdulaney 6y ago
If Bergstrom's point is that one shouldn't believe claims just because someone asserts them, then yes, that's true. But I would say that the set of assertions for which one is justified in rejecting them out of hand is perhaps smaller than he is suggesting. For example, there was fake news recently in which the Gates Foundation was said to be inserting subcutaneous chips into folks who had received vaccines. It turned out to be untrue, but not obviously so. The technology exists and if not abused it could be an inexpensive and effective form of record keeping.
- a1369209993 6y ago
  > rejecting them out of hand I may have missed it, but I don't see TFA actually arguing for that at all. > one shouldn't believe claims just because someone [or something] asserts them Seems to be the overall point. If some guy rolls in with a paranoid schizophrenic telling you someone is a criminal using facial features, you need to be able to say that is nuts.