Stephan (Martian Moonshine)

Karma: 43

I am a researcher in AI Interpretability at the Zuse Institute in Berlin.

Mostly interested in building theoretical foundation for Interpretability that works even if the agents have an incentive to hide their interpretations.

Stephan (Martian Moonshine) 15 Sep 2026 11:44 UTC
8 points
0
in reply to: Pretentious Penguin’s comment on: Maxime Fournes’s Shortform
That sounds reasonable. Every government should be briefed, but this seems of outsized importance.

Do you have the time to reach out to the AIS Norway chapter? Maybe these guys: https://aisnorway.github.io/aisnorway.org/en/contact/

Stephan (Martian Moonshine) 14 Sep 2026 18:15 UTC
31 points
23
in reply to: phoenix’s comment on: Maxime Fournes’s Shortform
I all seriousness, we should get the Nobel peace prize committee talking about AI safety. The Nobel is something Trump actually wants (and deserves if he acts on AI Safety). And he has only two more years to do something Nobel-worthy.

I think it’s not at all on their radar atm, at least there is nothing on their website.

https://www.nobelpeaceprize.org/research/what-we-do/nobel-peace-prize-forum/

We should get their attention!