Army general says he's using AI to improve "decision-making" (arstechnica.com)

The military using AI for decisions was one of the inflection points in AI 2027

https://www.youtube.com/watch?v=k_onqn68GHY

https://ai-2027.com/

I would bet my paycheck General Taylor is experiencing AI psychosis. AI is good for a lot of things, but military decision making isn’t one of them.

He could save power by just using dice and decision tables. There's a long history of civilian and military wargaming that's random but designed vs. real-world situations. Statistical token generation, however, doesn't strike me as a good bet. Hell, during one of the baseball games, Google AI led off their search results with "Seattle is not north of Toronto". It followed by having little real knowledge of historical baseball results. I can see it having problem due to training set with the latter, but damn, where countries and cities are located should be well-defined.

Sounds tricky if the LLM isn’t fine tuned to a dataset with examples of the kind of strategizing that you’d like to optimize for. Untrained LLMs are experts at average responses, and better strategies for anything require creativity, domain knowledge, and lateral thinking.