Nicholas Christakis

So you listen to those guys and you're like, well, they're also right.

Well, they can't both be right.

And you know, that's also true.

So I have sort of stopped trying to form in my own, because I'm not so expert in this area,

But I am expert in another area, which is related to this, which is this issue of how AI is going to change human behavior.

And here, just to preface one set of ideas, the kind of toy model that I like to throw out there to sort of help people fix ideas is imagine the manufacturer of an Alexa digital assistant.

The manufacturer of a digital assistant is very concerned with the human-machine interaction.

You would never buy an Alexa if every time you had to speak to it, you had to say, excuse me, Alexa, I'm very sorry to interrupt you.

If you don't mind, would you please tell me the weather tomorrow?

That would be an absurd level of politeness.

You would never buy a machine like that.

You expect to be able to say, Alexa, weather, and it obediently responds.

And that's fine until you bring the machine into your home and your children, in speaking to that machine, learn to be rude.

And then they go to the playground and they are rude to other children.

So what we've been studying in my lab is human-human interactions in the presence of machines.

And specifically, what we've been focusing on is little perturbations in the AI systems, in the machine systems, that modify how the humans interact with each other.

And in fact, what we're working on is not so much super smart AI to replace human cognition, but dumb AI to supplement human interaction.

And because the humans are smart, you can think of the AI as a kind of catalyst, like platinum in an organic chemistry reaction, that just facilitates the interaction of humans and helps optimize them.

And we've done a broad set of experiments that have shown this is possible, that you can improve human collective and individual performance through the thoughtful injection of AI agents into social systems.

We haven't looked at that specifically.