Hacker News Reader

This sounds kinda fake to me. Like, how did the AI have a concept of an operator, or the operator's physical location, or comms equipment used to communicate with the operator, and how did it game out the consequences of destroying the operator or comms equipment? It would need an extremely sophisticated model of the world that's well beyond anything GPT-4 evidences.

I'd guess the "AI" was another human in a wargame, not an actual AI.

View on HN

Reinforcement learning with a dynamic world model. It's a pretty standard approach to training agents, although there are a lot of problems with this technique as well. People have used it to create agents for Minecraft, racing games, Nethack, lots of old arcade games and so forth.

Turns out it's literally exactly what I said. A human came up with the idea. No AI involved: https://www.usatoday.com/story/news/politics/2023/06/02/us-a...

Reply on news.ycombinator.com