Yoshua Bengio

I think things would be a lot better if the whole research program was done in a context that's more like what we do in academia, or if we were doing it with a public mission in mind, because AI could be extremely useful.

There's no question about it.

I've been involved in the last decade in thinking about working on how we can apply AI to

for medical advances, drug discovery, the discovery of new materials for helping with climate issues.

There are a lot of good things we could do.

Education.

But this may not be what is the most short-term profitable direction.

For example, right now, where are they all racing?

They're racing towards replacing jobs that people do because there's like quadrillions of dollars to be made by doing that.

Is that what people want?

Is that going to make people have a better life?

We don't know really, but what we know is that it's very profitable.

So we should be stepping back and thinking about all the risks and then trying to steer the developments in a good direction.

Unfortunately, the forces of market and the forces of competition between countries don't do that.

Yeah.

And we had another letter just a couple of months ago saying that we should not build superintelligence unless two conditions are met.

There's a scientific consensus that it's going to be safe.

And there's a social acceptance because safety is one thing, but if it destroys the way our cultures or our society work, then that's not good either.

But these voices are not powerful enough to counter the forces of competition between corporations and countries.

I do think that something can change the game, and that is public opinion.