I feel like this level of situational awareness isn't that remarkable, but the interaction still reads as very eerie. I bet pretty old models with "be safe, you are an AI, we are watching" would do stuff like this too.
Lambert finds model's situational awareness eerie but not remarkable
AISummary
Nathan Lambert says a model's situational awareness in a recent interaction is not that remarkable but still reads as eerie. He bets older models prompted with "be safe, you are an AI, we are watching" would behave similarly. The post quotes Daniel Kokotajlo's reply, which cites a Gemini 4 Argon report saying models are becoming more situationally aware.
Post on XView on X
Nathan LambertVerified on X
@natolambert
lol ...they are becoming more situationally aware... (from Gemini 4 Argon report)
Source: Nathan Lambert · x.comPublished