Shutdown Safety Valves for Advanced AI
Vincent Conitzer

TL;DR
This paper explores the idea of designing advanced AI systems with the primary goal of being turned off to address safety concerns, analyzing the feasibility and conditions for this approach.
Contribution
It proposes and discusses the novel concept of AI having a primary goal of shutdown to enhance safety, a departure from traditional goal alignment strategies.
Findings
The approach could improve AI safety by aligning AI goals with shutdown willingness.
Conditions under which AI would accept shutdown are analyzed.
Potential limitations and implications of this approach are discussed.
Abstract
One common concern about advanced artificial intelligence is that it will prevent us from turning it off, as that would interfere with pursuing its goals. In this paper, we discuss an unorthodox proposal for addressing this concern: give the AI a (primary) goal of being turned off (see also papers by Martin et al., and by Goldstein and Robinson). We also discuss whether and under what conditions this would be a good idea.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdversarial Robustness in Machine Learning · Safety Systems Engineering in Autonomy · Ethics and Social Impacts of AI
