Cassie Kozyrkov

That's an important difference, and it's not a difference of the mathematics on the inside.

It's a difference in how we lead.

So there are decision questions all the way through these technologies, and these technologies will affect how we make decisions on the other side.

Please answer this one question.

Actually, I want to say something else.

To make these bold predictions about a thing, and I want to empower everyone here to make their own bold predictions, so I don't have to get stuck with having to be everyone's Nostradamus, despite being a recovering statistician, and that's what we do.

But let's start with the basics.

There are different kinds of AI all building on one another.

The first kind is the research kind, the mathematics.

In a nutshell, what's that?

That's finding patterns in data.

Do you care?

Not particularly.

That's fine.

Next one up from that.

If we have mathematics that can pull patterns out of data, what will we apply it to?

We begin to apply it to using examples instead of instructions to get our wishes across, right, at scale.

So you might say, well, what are the instructions that I would have to write to take a photograph and figure out what to do with every single pixel in there to classify whether the animal in it is a cat or is not a cat?

What's that formula?

Tell me, what are we doing with the top left pixel and the one next to it?