Elon Musk

And my opinion on AI safety is the most important thing is that it be maximally truth-seeking, like that you don't force the AI to believe things that are false.

And we've obviously seen some concerning things with AI that we've talked about, where Google Gemini, when it came out with the ImageGen,

And people said, like, you know, make an image of the founding fathers of the United States, and it was a group of diverse women.

Now, that is just a factually untrue thing.

And the AI knows it's factually untrue, but it's also being told that it has to...

Everything has to be deposed for a moment.

Now, the problem with that is that it can drive AI crazy.

You're telling AI to believe a lie, and that can have very disastrous consequences.

As it scales.

Yeah, let's say, like, if you've told the AI that diversity is the most important thing, and now assume that that becomes omnipotent, and you've also told it that there's nothing worse than misgendering.

So at one point, Char GVT and Gemini, if you asked which is worse, misgendering Caitlyn Jenner or global thermonuclear war where everyone dies, it would say misgendering Caitlyn Jenner.

which even Caitlyn Jenner disagrees with.

I think people don't quite appreciate the level of danger that we're in from the woke mind virus being effectively programmed into AI.

Because imagine as that AI gets more and more powerful, if it says the most important thing is diversity, the most important thing is no misgendering.

And then it will say, well, in order to ensure that no one gets misgendered, then if you eliminate all humans, then no one can get misgendered because there's no humans to do the misgendering.

So you can get in these very dystopian situations.

Or if it says that everyone must be diverse, it means that there can be no straight white men.

And so then you and I would get executed by the AI.

Yeah, because we're not in the picture.

You know, Gemini was asked to create a, you know, show an image of the Pope, once again, a diverse woman.