Absolutist AI

TL;DR

This paper proposes training AI with absolute constraints to improve safety, corrigibility, and environmental exploration, formalizing the concept and analyzing its implications for AI behavior and alignment.

Contribution

It introduces a decision-theoretic formalization of absolute constraints and analyzes their effects on AI safety, corrigibility, and behavior, offering a novel approach to alignment.

Findings

Absolutist AIs avoid irrational behavior and environmental pressure to maximize expected value.

Absolute constraints serve as effective safety guardrails and improve corrigibility.

Such AIs do not necessarily maximize expected value, reducing certain risks.

Abstract

This paper argues that training AI systems with absolute constraints -- which forbid certain acts irrespective of the amount of value they might produce -- may make considerable progress on many AI safety problems in principle. First, it provides a guardrail for avoiding the very worst outcomes of misalignment. Second, it could prevent AIs from causing catastrophes for the sake of very valuable consequences, such as replacing humans with a much larger number of beings living at a higher welfare level. Third, it makes systems more corrigible, allowing creators to make corrective interventions in them, such as altering their objective functions or shutting them down. And fourth, it helps systems explore their environment more safely by prohibiting them from exploring especially dangerous acts. I offer a decision-theoretic formalization of an absolute constraints, improving on existing models in the literature, and use this model to prove some results about the training and behavior of absolutist AIs. I conclude by showing that, although absolutist AIs will not maximize expected value, they will not be susceptible to behave irrationally, and they will not (contra coherence arguments) face environmental pressure to become expected-value maximizers.