Learning norms from stories: A prior for value aligned agents

Md Sultan Al Nahian, Spencer Frazier, Mark Riedl, Brent Harrison

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

7 Scopus citations

Abstract

Value alignment is a property of an intelligent agent indicating that it can only pursue goals and activities that are beneficial to humans. Traditional approaches to value alignment use imitation learning or preference learning to infer the values of humans by observing their behavior.We introduce a complementary technique in which a value-aligned prior is learned from naturally occurring stories which encode societal norms. Training data is sourced from the children's educational comic strip, Goofus & Gallant. In this work, we train multiple machine learning models to classify natural language descriptions of situations found in the comic strip as normative or non-normative by identifying if they align with the main characters' behavior. We also report the models' performance when transferring to two unrelated tasks with little to no additional training on the new task.

Original languageEnglish
Title of host publicationAIES 2020 - Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
Pages124-130
Number of pages7
ISBN (Electronic)9781450371100
DOIs
StatePublished - Feb 7 2020
Event3rd AAAI/ACM Conference on AI, Ethics, and Society, AIES 2020, co-located with AAAI 2020 - New York, United States
Duration: Feb 7 2020Feb 8 2020

Publication series

NameAIES 2020 - Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

Conference

Conference3rd AAAI/ACM Conference on AI, Ethics, and Society, AIES 2020, co-located with AAAI 2020
Country/TerritoryUnited States
CityNew York
Period2/7/202/8/20

Bibliographical note

Funding Information:
This material is based upon work supported by the National Science Foundation under Grant No. 1849231.

Publisher Copyright:
© 2020 Copyright held by the owner/author(s).

Keywords

  • Learning from Stories
  • Natural Language Processing
  • Value alignment

ASJC Scopus subject areas

  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Learning norms from stories: A prior for value aligned agents'. Together they form a unique fingerprint.

Cite this