You're listening to American Prestige. To listen ad-free, you can subscribe at Americanprostagepod.com. Find the link in our show notes.

Defense policy, but given that we're in the middle of war and AI has become suddenly a big issue, we're switching focus a little bit to talk about tech and the Iran

more specifically. So let's just start with what we know in terms of how the defense department

is using so-called AI LLMs in its targeting packages or any other aspect of the conflict.

Sure. So I mean, fortunately or unfortunately, who the fuck is Alkscarp is still a very relevant question and answer here.

Alexcarp's company, Pallantier, has a product called the Maven Smart System, which is the sort of grandchild.

People might remember Project Maven, the controversy at Google from about eight years back.

It has turned into this Palantir tool that is now used by the Department of Defense to ingest all sorts of disparate, uh, data feeds, you know, drone footage, satellite footage,

intelligence reports, you know, stuff from spy agencies, et cetera, et cetera, et cetera,

cramble that together and then now apply an LLM on top of it to decide, okay, we had this

comprehensive picture of what's on the ground. Let's ask Claude what to blow up first, is essentially what it boils down to.

So, I mean, that much is in the public record.

What's been reported recently vis-à-vis Iran is that the DOD is using, or sorry, DOW is using

Anthropics Claude chatbot

to basically to speed up

the airstrike process

from picking a target to

killing people, you know, blowing things up on the ground.

This, you know, how exactly they are querying Claude in this case is a little unclear.

But I think the most important thing that's been reported is that it is speed is a factor.

This is being used to do airstrikes faster to pick targets faster.

I mean, Claude will literally suggest things for sentcom to attack.

...