mlsanityData to 14 Oct 2026

Off-Switching Not Guaranteed

arXiv 2502.08864 ↗First posted 13 Feb 2025cs.AIcited by 4

Not on the map. Why?

Abstract (arXiv)

Hadfield-Menell et al. (2017) propose the Off-Switch Game, a model of Human-AI cooperation in which AI agents always defer to humans because they are uncertain about our preferences. I explain two reasons why AI agents might not defer. First, AI agents might not value learning. Second, even if AI agents value learning, they might not be certain to learn our actual preferences.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 6 of its 43 references that are in the map. First posted on arXiv 13 Feb 2025.

Builds on

6 of 43 references.

References · 6

  1. 2016Cooperative Inverse Reinforcement Learningcited by 300
  2. 2016The Off-Switch Gamecited by 97
  3. 2023Credal Bayesian Deep Learningcited by 45
  4. 2017Should Robots be Obedient?cited by 20
  5. 2017Incorrigibility in the CIRL Frameworkcited by 9
  6. 2023Rational Aversion to Informationcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 3.

4 papers.

  1. 2026Against Proxy Optimizationcited by 1
  2. 2025AI Agents and Hard Choicescited by 0
  3. 2026Shutdown Safety Valves for Advanced AIcited by 0
  4. 2026Revisiting the shutdown problemcited by 0