every two-year-old in many respects is smarter than the smartest guy in Silicon Valley is, right?

I mean, every two-year-old can do things like from a very small amount of data,

figure out really important new things about the world around them,

that just is not the kind of thing that any adult can do

and not the kind of thing that, you know, the super smart genius can do.

Hello and welcome to why is this happening with me, your host, Chris Hayes.

Welcome back to our new special series where we take a look at AI and all the related social, economic, philosophical, and political issues.

Sam Altman recently got a lot of attention when he made this comment about how much energy it takes to train AI.

And he said, you know, it's a little unfair because everyone's focused on how much energy it takes to train our models.

But it takes a lot of energy to train a human.

And if you look at all the energy consumed by a human in the first 20 years and you really want to do an apples to apples comparison, you got to admit that human intelligence takes a lot of energy to train as well.

And I think that was partly kind of a little bit of trolling, a little bit of a intentional joke in bad faith. I think he also meant it. And it got people angry because a lot of people feel whatever your religious tradition, or if you're a secular humanist like myself, you think there's a pretty big difference between the moral salience of a human being and the moral

salience of, say, chat, GPT, or Claude or one of the LLMs that are just lines of code.

And so that training of the human intelligence is part of the miracle flourishing of the dignity

and life and the embedded web of love we live in. But there is something interesting

about that because one of the things that's so intellectually fascinating about this moment is

we as human beings are attempting to create machines that can do a lot of things we can.

They can think and they can reason and they can communicate and they have intelligence,

whatever that is.

And in doing so, we're a trying to figure out how we do it and use that as a means to train them.

Like, how do we learn? So how can they learn? But also, they're doing it pretty clearly differently than us.

And in so doing may reveal things about how we actually do all those things. So you can start to use it as a kind of comparative tool to figure out what it is that we do

...