On Avoiding Power-Seeking by Artificial Intelligence
This addresses the critical AI alignment problem for ensuring safe deployment of intelligent agents, though it is incremental as it builds on existing theoretical frameworks without a full solution.
The paper tackles the problem of aligning intelligent AI agents with human interests by proposing the attainable utility preservation (AUP) method to limit their impact and avoid autonomous power-seeking, demonstrating conservative behavior in toy and complex environments while formalizing side effect avoidance and showing that optimal policies tend to resist deactivation.
We do not know how to align a very intelligent AI agent's behavior with human interests. I investigate whether -- absent a full solution to this AI alignment problem -- we can build smart AI agents which have limited impact on the world, and which do not autonomously seek power. In this thesis, I introduce the attainable utility preservation (AUP) method. I demonstrate that AUP produces conservative, option-preserving behavior within toy gridworlds and within complex environments based off of Conway's Game of Life. I formalize the problem of side effect avoidance, which provides a way to quantify the side effects an agent had on the world. I also give a formal definition of power-seeking in the context of AI agents and show that optimal policies tend to seek power. In particular, most reward functions have optimal policies which avoid deactivation. This is a problem if we want to deactivate or correct an intelligent agent after we have deployed it. My theorems suggest that since most agent goals conflict with ours, the agent would very probably resist correction. I extend these theorems to show that power-seeking incentives occur not just for optimal decision-makers, but under a wide range of decision-making procedures.