Yoshua Bengio

We're not playing with the atmosphere to try to fix climate change because we might create more harm than actually fixing the problem.

We are not creating new forms of life that could destroy us all, even though it's something that is now conceived by biologists because the risks are so huge.

But in AI...

It isn't what's currently happening.

We're taking crazy risks.

But the important point here is that even if it was only a 1% probability, let's say, just to give a number, even that would be unbearable, would be unacceptable.

Like a 1% probability that our world disappears, that humanity disappears, or that a worldwide dictator takes over thanks to AI.

These sorts of scenarios are so catastrophic.

that even if it was 0.1%, it would still be unbearable.

And in many polls, for example, of machine learning researchers, the people who are building these things, the numbers are much higher.

We're talking more like 10% or something of that order, which means we should be just paying a whole lot more attention to this than we currently are as a society.

Experts disagree that

And they range in their estimates of how likely it's going to be from tiny to 99%.

So that's a very large bracket.

Let's say I'm not a scientist, and I hear the experts disagree among each other.

And some of them say it's very likely, and some say, well, maybe it's plausible, 10%.

And others say, oh, no, it's impossible, or it's so small.

Well, what does that mean?

It means that we don't have enough information to know what's going to happen, but it is plausible that one of the more pessimistic people in the lot are right, because there is no argument that either side has found to deny the possibility.

I don't know of any other existential threat that we could do something about that has these characteristics.