h/t to Ed Zitron: https://bsky.app/profile/edzitron.com/post/3mfxqjqoias2q
alt text
WSJ PATRICK SISON/ASSOCIATED PRESS Within hours of declaring that the federal government will end its use of artificial-intelligence tools made by tech company Anthropic, President Trump launched a major air attack in Iran with the help of those very same tools. Commands around the world, including U.S. Central Command in the Middle East, use Anthropic’s Claude AI tool, people familiar with the matter confirmed. Centcom declined to comment about specific systems being used in its ongoing operation against Iran. The command uses the tool for intelligence assessments, target identification and simulating battle scenarios even as tension between the company and Pentagon ratcheted up, the people said, highlighting how embedded the AI tools are in military operations. The administration and Anthropic have been feuding for months over how its AI models can be used by the Pentagon. Trump on Friday ordered agencies to stop working with the company and the Defense Department designated it a security threat and risk to its supply chain.
Simulated battle scenarios are a common component of wargaming. That doesn’t mean an LLM is the right tool for it, but it’s been a thing for a long time.
The bigger concern here is using it for intelligence assessments and target acquisition, because LLMs hallucinate a lot.
as a side effect, it’s a phenomenal accountability sink. people almost forget that usaf can make entirely human-made fuckups https://en.wikipedia.org/wiki/Amiriyah_shelter_bombing
Don’t forget that time we leveled a clearly-marked hospital that we were in radio contact with the entire time.
What about the time the US carpet bombed an entire company of Canadian soldiers in iraq?
Yeah, now when your autonomous weapon systems target your own fighter jets, no one gets court martialled!
@wonderingwanderer @fullsquare TBF, fighter jets should have been unmanned drones
On the one hand, an autonomous fighter jet would be immune to G-LOC, letting them perform maneuvers that would incapacitate/kill a human pilot. On the other hand, air-to-air combat is a complex affair, and the enemy will be probing for any weaknesses in your drones’ programming to exploit.
Autonomous bombers seem easier to pull off - bombing missions are (relatively) straightforward compared to air-to-air combat.
That’s fair, I only meant it to poke fun at the LLM simulating battle scenarios, I know it’s useful in general to simulate and wargame
Yeah, an LLM is not designed for those kinds of simulations. It can write you a choose-your-own adventure story, but it can’t realistically model dynamic kinetic operations with any degree of applicability.