Cassie Kozyrkov

The rest of you, click.

I show them a landmine.

And I ask them, how is this different from an AI system in terms of the decision making?

How is it different?

Does it make decisions?

Well, it responds to stimuli.

It responds to inputs.

It collects data in some sense in terms of pressure.

And it is also a system whose effect is fairly complex relative to the understanding of whoever put it in the ground.

Like they didn't know what was going to happen and why and how it was going to be triggered.

But still though, so it has a lingering effect, it has a potentially powerful effect, complex, hard to predict.

But it's not autonomous in the sense that humans can wash their hands of responsibility for it.

And it's not autonomous in the sense that no human actually made it, thought it was a good idea to create them, thought it was a good idea to put it in the ground, right?

It is still a thing of human responsibility.

So when we build these systems, when we integrate these systems,

We have to be careful and wise about how we do that.

All of it is human decisions.

There are good ways and there are bad ways.

You could build systems with safety nets.

You could build them without safety nets.