If you think about a general intelligence, an artificial general intelligence, an AI that is as versatile as a human, and then you put a voice on it and you embody it in, say, a human shape–then you create a robot with a general intelligence. We all can kind of imagine what that would be like. Now, can we build such a device? We don’t know know. At an intuitive level, people seem to think, yes, but that’s mainly, I believe, because you see that in movies. You see things like Westworld, where you have a city of robots that are indistinguishable from humans, and, of course, are played by humans. You think of Ex Machina, you think of Her which is Scarlett Johansson voicing a robot.
But using those as kind of intuitive proof points is a logical fallacy called, “generalizing from fictional evidence.” It’s something we all do. We’ve seen that so many times that it seems reasonable. But if you take a step back and say, “What evidence do we actually have that we can build such a thing?” and you’re forced back into the real world, looking at the kinds of systems we have now, they don’t resemble those at all. So one wonders if we aren’t taken in, as it were, by the movie, by the TV show, or in fact those are telling us what we can in fact build and what we will be building soon.