Using stories to teach human values to artificial agents

Mark O. Riedl, Brent Harrison

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

44 Scopus citations

Abstract

Value alignment is a property of an intelligent agent indicating that it can only pursue goals that are beneficial to humans. Successful value alignment should ensure that an artificial general intelligence cannot intentionally or unintentionally perform behaviors that adversely affect humans. This is problematic in practice since it is difficult to exhaustively enumerated by human programmers. In order for successful value alignment, we argue that values should be learned. In this paper, we hypothesize that an artificial intelligence that can read and understand stories can learn the values tacitly held by the culture from which the stories originate. We describe preliminary work on using stories to generate a value-aligned reward signal for reinforcement learning agents that prevents psychotic-appearing behavior.

Original languageEnglish
Title of host publicationWS-16-01
Subtitle of host publicationArtificial Intelligence Applied to Assistive Technologies and Smart Environments; WS-16-02: AI, Ethics, and Society; WS-16-03: Artificial Intelligence for Cyber Security; WS-16-04: Artificial Intelligence for Smart Grids and Smart Buildings; WS-16-05: Beyond NP; WS-16-06: Computer Poker and Imperfect Information Games; WS-16-07: Declarative Learning Based Programming; WS-16-08: Expanding the Boundaries of Health Informatics Using AI; WS-16-09: Incentives and Trust in Electronic Communities; WS-16-10: Knowledge Extraction from Text; WS-16-11: Multiagent Interaction without Prior Coordination; WS-16-12: Planning for Hybrid Systems; WS-16-13: Scholarly Big Data: AI Perspectives, Challenges, and Ideas; WS-16-14: Symbiotic Cognitive Systems; WS-16-15: World Wide Web and Population Health Intelligence
Pages105-112
Number of pages8
ISBN (Electronic)9781577357599
StatePublished - 2016
Event30th AAAI Conference on Artificial Intelligence, AAAI 2016 - Phoenix, United States
Duration: Feb 12 2016Feb 17 2016

Publication series

NameAAAI Workshop - Technical Report
VolumeWS-16-01 - WS-16-15

Conference

Conference30th AAAI Conference on Artificial Intelligence, AAAI 2016
Country/TerritoryUnited States
CityPhoenix
Period2/12/162/17/16

Bibliographical note

Publisher Copyright:
Copyright © 2016, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.

ASJC Scopus subject areas

  • General Engineering

Fingerprint

Dive into the research topics of 'Using stories to teach human values to artificial agents'. Together they form a unique fingerprint.

Cite this