Abstract

At the recent AGI-11 Conference Orseau and Ring, and Dewey, described problems, including self-delusion, with the behavior of AIXI agents using various definitions of utility functions. An agent's utility function is defined in terms of the agent's history of interactions with its environment. This paper argues that the behavior problems can be avoided by formulating the utility function in two steps: 1) inferring a model of the environment from interactions, and 2) computing utility as a function of the environment model. The paper also argues that agents will not choose to modify their utility functions.

Description

Model-based Utility Functions

Links and resources

Tags