Humans in the Loop
Dr. Gary Shiffman, a behavioral economist and the former chief of staff for U.S. Customs and Border Protection, is the CEO of the growing analytics startup Giant Oak, one of the contractors for ICE’s Homeland Security Investigations division that I recently wrote about in The New York Times Magazine. “We see the people behind the data,” Giant Oak’s website reads. Shiffman emailed after the story was published to say he disagrees with my addendum to something he said. The passage in question:
[Shiffman] made clear that he didn’t want to live in a country that indiscriminately rounded up foreigners. But if public safety was the goal, the answer wasn’t to stop scanning people’s data. It was to do it the right way. “Now this is my theory,” he told me, “and you tell me if it’s crazy or wrong. But I think that the better we have entity resolution” — that is, the better we can compile and measure people’s data — “the less of a surveillance state we’ll have.” Once the machine is perfect, as long as you’re following the rules, you won’t even have to know it’s there.
Giant Oak’s software, according to its website, can “screen thousands of people in a matter of minutes,” prioritizing and ranking the results so that governments or financial institutions can “focus first on targets whose behavior pose high decision risk.” While it may make human investigators faster, it doesn’t replace them. So if my addendum left readers with the impression that artificial general intelligence is Giant Oak’s product or goal, I agree that’s the wrong impression. The questions about semi-automated decision-making—about the loss of friction, about what data we collect and what data is relevant—are more subtle and interesting than whether we’re collectively building HAL 9000. (We aren’t, yet.)
Shiffman has a blog post out on this today: https://resources.giantoak.com/beware-of-perfect-machines
And spoke more about the fallacy of HAL 9000 in testimony before a House subcommittee earlier this year: http://blog.giantoak.com/blog/an-economists-view-on-technology-in-the-future-of-bsa/aml