Andrea Miotti

Good to see you, ma'am.

Thank you for having me on the show.

Yeah, so while this is a fairly innocuous scenario, this person just lost all of her emails.

It's annoying, but it's not catastrophic.

I think this should be a wake-up call for most people about where AI is going.

So what these AI companies are building is not just chatbots.

And I think things like Cloudbot has really shown the world that this is not the case.

They're building what they call superintelligence, which is AI systems that are meant to replace and outcompete humans across all tasks.

AI systems that can use computers, they can use tools, they can essentially do anything you can do from a computer, but faster and better.

And already right now, it's really hard to keep these very smart and very powerful AI systems under control.

Here, this meta exec, she couldn't get the system to not just delete her emails.

And the more these AI systems are made to be smarter by these AI companies, and they're integrated across the economy, the more we'll see more and more dangerous situations where they're not just deciding on email deletion, but they're deciding on

situations that could be life and death for people.

And ultimately, this is a very dangerous direction that the AI industry is going on to develop these AI systems that we don't know how to control.

They're making them explicitly better than humans across the board and with no plan in sight for how to keep them under human control and keep us safe.

Yeah, and I think the important thing is the Pentagon has all of its rights to select which provider it thinks are best.

But the issue is not which human is giving this order.

The issue is that these AI companies don't know how to control their own systems.

The issue we're going to face is one where no human

knows how to keep them under control.