‘The discourse is unhinged’: how the media gets AI alarmingly wrong – Info AI

The thing that I think people get wrong about AI is not what it can do but how it can do it. This is something that I see when people talk about animals, for example. They say something along the lines of “Did you know that animal X can do action Y? Isn’t that amazing!” What gets me is that they are imagining what it would take for a human, say themselves, to perform that action, and then assuming that this must be what the animal is going through in order to do it. Which could not be farther from the truth. The other thing they assume, and this somewhat goes hand in hand, is that if animal can perform action Y, and it is equally difficult for a human to perform action Y and action Z, then the animal should be able to also do action Z; again, nonsense of course. The same is true of AI. The mechanics of how they actually perform the tasks that they do are in reality relatively simplistic and useless except in heavily, heavily restricted domains. That’s the big picture of AI that the media misses.