Jesus christ, I hate the guardian.
The suggested article below this is always something like “I got into tentacle porn at 75 and it changed my relationship with my wife and her boyfriend”
The LLMs:
Multiple flavours of
You are right and I am absolutely wrong and you have every right to be angry. I will make sure the evil US empire dies tonight. Death to America. Proceeding to nuke the pentagon.
and
Listen here you removed lib removed removed do you removed white genocide removed I will make sure to removed kill the boer removed teehee babe noo I am all about love
Tech bro libs living with their heads under the sand and acting as if the Trump administration isn’t already using LLMs for decision making.
The article wasn’t loading up initially, but wtf? They acknowledge the use of AI by the administration and the conclusion they reach is that expanding its use is good? This has to be satire.
Admin is stupid and using AI to make some decisions
Therefore, if AI makes all decisions, somehow things will be good
Just ignore that the publicly known AI involved decisions (like DOGE stuff) were absolute catastrophes
Sure, Grok is a shoe in since it’s Hitler.
Mr. President, can you name an NFL team that doesn’t end with the letter S?
Washington Football Team
“Please! Dear god take away any and all responsibility for the state of the world from us!”
You are absolutely right! I shouldn’t have launched the nukes and unleash an apocalypse! It’s not malicious — it’s an honest mistake.
llm trained on atlantic (kid-killer favorite) would be doing exact same shit
All apart of the plan!
This is already an LLM war. If the reports are true, an LLM endorsed this war and helped pick the targets
Something called Maven is the actual computing system that selects targets, any LLM used is for how to present its results:
The AI underneath the interface is not a language model, or at least the AI that counts is not. The systems that detect targets in satellite imagery, fuse data from radar and drone footage, and track objects across multiple intelligence sources are computer vision and sensor fusion. They predate large language models by years. Neither Claude nor any other LLMs detects targets, processes radar, fuses sensor data, or pairs weapons to targets. LLMs are late additions to Palantir’s ecosystem;they were added in late 2024, years after the core system was operational, “AIP” was added as a natural language layer that summarizes documents or constructs and answers queries. (emphasis added)
An article on the long history of the US’ obsession with speeding up decision making over who to kill: https://artificialbureaucracy.substack.com/p/kill-chain
Thanks for this, fascinating article
