I really ought to know better than to expect sensible commentary on computer science from Wired, but their The Other Turing Test article crossed the line of agitation for me. I’m fine with them deciding to recount a project by some undergraduates to replicate the male/female imitation game scenario in the Turing Test, and adding ALICE into the mix is interesting, though I think it misreads Turing’s intention to say that the computer has to pretend to be female, as compared to just human. But to say: “Scientists studying artificial intelligence have long argued over the meaning of this gender-bending experiment” is just too much. If one reads Turing’s paper, Computing Machinery and Intelligence proposing the imitation game, one sees in the last paragraph of section 1 that the Turing Test poses the question of whether, when a person and computer play the imitation game, “Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman?”(emphasis mine) In other words, Turing is saying that computers do not have to be able to pass as more human than a human 100% of the time, or even 50% of the time. They just have to be as good at passing as something they are not as often as people are.
Now, there can be much interesting discussion of why Turing chose an all-human imitation baseline of determining gender as compared to other possible discriminators. But the gendered imitation game has a clear role in defining success in the imitation game. If it has been “ignored”, it is because there is little point in calculating the baseline before artificial intelligence programs realistically capable of passing the Turing Test exist. I’m not sure what an academic publication on the gendered version would really conribute to the field, anyway. And that is clearly the definition of “ignored” they are using, because many people (myself included) have held classroom versions of the gendered imitation game with our students.