First, we just get your reactions to what you have learned about DeepSeek in the last 24 hours.

How do you make of it? How could this change things?

Well, exactly what Andrew L. Torquins said, there's a lot of open questions.

And, you know, I think that the likelihood it was trained for as little money as it said, it seems very unlikely.

It also seems likely that they actually,

to train these smaller models, which we already knew in the US,

you have to have access to and use larger models,

like the Lama, chat GPT, et cetera.

So look, I think it's extremely important and interesting result.

I don't think that that suggests that the large models

are actually not essentially valuable in creating things like Deep Zick.

Okay, so let's turn now to what you're hoping to do. Tell us more about the good.

We hear a lot about the worries people have about AI, the sort of scary future that it could portend.

But its advocates say a lot of good can be done with it. Hard to imagine

something better caused and curing cancer. Talk to us about that. So, you know, as you mentioned,

the general discourse and actually human history, when a new major technology comes wrong,

whether it's the printing press, electricity, the car is, oh my God, everything's going to go

wrong, right? Like society's going to break, human agency is going to break, et cetera.

And yet, when we look at the end result, like we can't have the society without the printing

press, electricity, the computer, etc.

It turns out that it massively increases human agency.

And part of that increase of human agency is when we all get access to it,

...