Human Control: Definitions and Algorithms
This work addresses the critical issue of AI safety for researchers and policymakers, but it is incremental as it builds on existing concepts like corrigibility.
The paper tackles the problem of ensuring human control over advanced AI systems by formally defining shutdown instructability, a variant of corrigibility, and shows that it leads to appropriate shutdown behavior, retention of human autonomy, and avoidance of user harm.
How can humans stay in control of advanced artificial intelligence systems? One proposal is corrigibility, which requires the agent to follow the instructions of a human overseer, without inappropriately influencing them. In this paper, we formally define a variant of corrigibility called shutdown instructability, and show that it implies appropriate shutdown behavior, retention of human autonomy, and avoidance of user harm. We also analyse the related concepts of non-obstruction and shutdown alignment, three previously proposed algorithms for human control, and one new algorithm.