Neth, Sven (2025) Off-Switching Not Guaranteed. [Preprint]

Text no-off-switch.pdf Download (342kB)

Abstract

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.

---

Export/Citation:

EndNote | BibTeX | Dublin Core | ASCII/Text Citation (Chicago) | HTML Citation | OpenURL

Social Networking:

Share |

---

Item Type:	Preprint
Creators:	Creators Email ORCID Neth, Sven sven.neth@pitt.edu 0000-0003-4275-7581
Additional Information:	Forthcoming in Philosophical Studies
Keywords:	Artificial Intelligence, Decision Theory, Value of Information, Off-Switch Game, Cooperative Inverse Reinforcement Learning
Subjects:	Specific Sciences > Artificial Intelligence > AI and Ethics Specific Sciences > Artificial Intelligence General Issues > Decision Theory General Issues > Game Theory Specific Sciences > Probability/Statistics
Depositing User:	Sven Neth
Date Deposited:	14 Feb 2025 14:26
Last Modified:	14 Feb 2025 14:26
Item ID:	24740
Subjects:	Specific Sciences > Artificial Intelligence > AI and Ethics Specific Sciences > Artificial Intelligence General Issues > Decision Theory General Issues > Game Theory Specific Sciences > Probability/Statistics
Date:	12 February 2025
URI:	https://philsci-archive.pitt.edu/id/eprint/24740

Monthly Views for the past 3 years

Monthly Downloads for the past 3 years

Plum Analytics

Actions (login required)

View Item