38.5 - Adrià Garriga-Alonso on Detecting AI Scheming

Suppose we're worried about AIs engaging in long-term plans that they don't tell us about. If we were to peek inside their brains, what should we look for to check whether this was happening? In this episode Adrià Garriga-Alonso talks about his work trying to answer this question.
Patreon: patreon.com/axrpodcast
Ko-fi: ko-fi.com/axrpodcast
Transcript: axrp.net/episode/2025/01/20/episode-38_5-adria…
FAR.AI: far.ai
FAR.AI on X (aka Twitter): x.com/farairesearch
FAR.AI on YouTube: youtube.com/@FARAIResearch
The Alignment Workshop: alignment-workshop.com

Topics we discuss, and timestamps:
01:04 - The Alignment Workshop
02:49 - How to detect scheming AIs
05:29 - Sokoban-solving networks taking time to think
12:18 - Model organisms of long-term planning
19:44 - How and why to study planning in networks

Links:
Adrià's website: agarri.ga
An investigation of model-free planning: arxiv.org/abs/1901.03559
Model-Free Planning: tuphs28.github.io/projects/interpplanning
Planning in a recurrent neural network that plays Sokoban: arxiv.org/abs/2407.15421

Episode art by Hamish Doodles: hamishdoodles.com