So the AI can fool a lot of people … in the sense that it can pass for human often enough? 😊

Something like that. For specific purposes we don’t need AGI                                            

OK, so the robot can react appropriately when it knows enough of the topic

Right, so as a receptionist at a hotel a robot doesn’t need to know everything, we can predict most of the questions that come up

When is breakfast, where is my room, can you order a taxi, etc.

Exactly! So a robot that can move convincingly, speak convincingly, and then look really human  

“look really human”, con you specify that? 🤗

Smile, nod, move their eyes around, gesticulate, have non-plasticky skin, no gaps, hinges, wires, etc.

That’s a daunting list of do’s and don’ts, and I guess there is a lot more 🧐

Yep. But if we can do that, we can climb out of uncanny valley                             

So how do we do that? 🤔

Well, we did manage to tackle a large part of speech recognition and production …             

… sure, fair enough, Siri and Alexa do work pretty well 😊

And we did that mostly by just studying humans and copying them 😊

Right, to pass the Turing test, imitate a human, fool the judge

Correct. So now we need to do the same thing with the rest, both body and behavior            

That would be a huge deal! 😁

Absolutely. And this is really necessary for empathy.                              

Tell me more!  😁