Gridworld agents trained with DReST reward functions learn to be USEFUL at tasks conditional on trajectory length and NEUTRAL across lengths, supplying initial evidence that the method could produce shutdownable agents.
'Indifference' methods for managing agent rewards
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
`Indifference' refers to a class of methods used to control reward based agents. Indifference techniques aim to achieve one or more of three distinct goals: rewards dependent on certain events (without the agent being motivated to manipulate the probability of those events), effective disbelief (where agents behave as if particular events could never happen), and seamless transition from one reward function to another (with the agent acting as if this change is unanticipated). This paper presents several methods for achieving these goals in the POMDP setting, establishing their uses, strengths, and requirements. These methods of control work even when the implications of the agent's reward are otherwise not fully understood.
fields
cs.AI 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Shutdownable Agents via Stochastic Choice
Gridworld agents trained with DReST reward functions learn to be USEFUL at tasks conditional on trajectory length and NEUTRAL across lengths, supplying initial evidence that the method could produce shutdownable agents.