Comment Re:I read methods... this is why. (Score 1) 212
This. The study is *bad*. Really bad.
Seriously, they say outright, the pay also maps directly to which AI assistant was perceived as more human. With a sample size of one each, to just declare that the male and female presenting avatar are equally believable, and it must be gender that's the cause is a . . . choice.
But I really can't see how they can possibly defend it short of, they like the result this way, it's a good headline.
But that's literally what they did. They tested about 100 people who with one set of assistants. The same people consistently said one was more lifelike than the other. They concluded, for reasons, that must be because of gender. Completely ridiculous. The only reason anyone should do anything other than ridicule this, is because they're scared of the politics of saying so. But it is bad.