Joe Allen

No, it's determined because the society itself as it's influenced and controlled by these tech oligarchs

has been refashioned so that only those people who have adopted, those people who have submitted, those people who have, in essence, merged, are able to remain relevant.

It is adaptation, but it's adaptation to an artificial ecosystem, a man-made ecosystem that was developed not necessarily, and I would say definitely not, for the benefit of those organisms

underneath the top of the food chain, those organisms being us.

No, this artificial ecosystem was developed solely for the benefit of those at the top and their ambitions to create first widely deployed artificial narrow intelligences and then to develop artificial general and super intelligence.

Their ambitions are clear.

They don't simply want to upgrade humans with effective algorithms.

The ultimate intention is to create digital gods.

Perhaps if one company rockets ahead with super intelligence, a single digital god, and they want you to submit to it.

That choice will be yours.

But before we get too crazy, I want to welcome

Our guest, who has just come off of Capitol Hill, where he was explaining to politicians and various government agents what the real dangers of artificial intelligence are.

I'm pleased to welcome Jeffrey Ladish from the Palisade Research Institute.

Am I saying this right?

Jeffrey is now well known.

We've seen a lot of media coverage of Palisade's work in The Guardian, New York Times, and elsewhere.

But Jeffrey works on evaluations.

His job is to basically engage with these AI models from Grok to OpenAI to Claude and determine exactly how smart they are.

So Jeffrey, I'd like to begin.

with just a breakdown of probably the most well-known Palisade study in which you had models trying to evade a shutdown.