From NBC News in Washington, the longest running show in television history.

This is Meet the Press with Chuck Todd.

Hello from Washington. I'm Chuck Todd with a special edition of Meet the Press.

This Sunday, we're bringing you our best interviews on artificial intelligence.

As AI technologies improve and creep deeper into our lives by all indications,

AI artificial intelligence is getting exponentially more intelligent and powerful. It has the

potential to do good, like revolutionizing cancer detection. But it also poses great perils just

this month. Hundreds of top industry leaders append a short letter warning that unfettered

AI could lead to

human extinction. In the next hour, we're going to give you the good, the bad, and the scary when it

comes to AI. And I want to begin with J.C. Reese Anthus, who argues we need an AI rights movement.

And I started off by asking him to set the stage, as America seems caught up in its how moment, to reference 2001 a space odyssey, with the engineering of AI ahead of the science and our ability to grasp it.

Yeah, there are maybe three ways we can build things. We can engineer them, as you said. We can do them scientifically. If you hear about the olden days of AI in the late 1900s, it was about building

expert systems, rule-based systems that took everything humans knew and went through kind of a

flow chart. We really understood from their ground up how they worked. Engineering is another level

where you have a high-level understanding, you bring some parts together. I don't even know if AI is

there. What we're doing now looks much more like growing, living things. It's like gardening or raising an animal or even a human child, you know, a superhuman, very strange human child. But we put in inputs to the process. We give it data. We say, here's what the world is like. We give it some rules for learning from that. So we say, when you get a new data point, here's how you predict some outcome and change

the way you think about the world. So you can predict the outcome, next outcome better.

And that leads to all sorts of strange systems. And there's a whole, there is a science of what

we call deep learning, a science of these big AI models, but it's

light years behind the engineering and the growing and incubation side.

But the bigger conclusion I've come to over the last couple of weeks is that it's too late to do

...