AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Parametrically Retargetable Decision-Makers Tend To Seek Power

Alexander Matt Turner, Prasad Tadepalli

NeurIPS 2022 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Human-AI coordination agents · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use a decision procedure that can be retargeted from staying in the initial room to leaving itGreedy final-observation reward maximization in Montezuma's Revenge · Montezuma's Revenge, step-T observationpreference for leaving over stayingImproved
Use ϵ-greedy exploration to learn an exploitation policyBandit algorithm for Montezuma's Revenge initial actions · Montezuma's Revenge initial actionretargetability multiplicity from ↓ to other actions: 4Improved
Increase training trials so the learned exploitation policy is more likely to try every actionϵ-greedy bandit algorithm · Montezuma's Revenge initial actionapproximation to greedy reward maximizationImproved
Use DQN rather than a training process capable of exploring beyond the first roomMontezuma's Revenge policy training · Leaving the first roomretargetability away from the first roomHurt
Enable exploration of the first three rooms to the right of the initial roomRL training with featurized item rewards · Montezuma's Revenge, leaving the first roomretargetability away from the first roomImproved
Increase exploration and ability to remember routes to item configurationsGO-EXPLORE-style training · Montezuma's Revenge, entire levelretargetability across final inventory outcomesImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy