The Effect Of Agent Reasoning Transparency On Automation Bias: An Analysis Of Response Performance

Keywords

Agent transparency; Automation bias; Complacency; Human-agent teaming

Abstract

We examined how the transparency of an agent’s reasoning affected the human operator’s complacent behavior in a military route selection task. Participants guided a three-vehicle convoy through a simulated environment in which they had a limited amount of information about their surroundings, all this while maintaining communication with command and monitoring their surroundings for threats. The intelligent route-planning agent, RoboLeader, assessed potential threats and offered changes to the planned route as necessary. RoboLeader reliability was 66 %, and the participant had to correctly reject RoboLeader’s suggestion when incorrect. Access to RoboLeader’s reasoning was varied across three conditions (no reasoning, reasoning present, and increased reasoning transparency), and each participant was assigned to one of the three conditions. Access to agent reasoning improved performance and decreased automation bias. However, when reasoning transparency increased, performance decreased while automation bias increased. Implications for presentation of reasoning information in operational settings are discussed in light of these findings.

Publication Date

1-1-2016

Publication Title

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Volume

9740

Number of Pages

465-477

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

DOI Link

https://doi.org/10.1007/978-3-319-39907-2_45

Socpus ID

84978795768 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84978795768

This document is currently not available here.

Share

COinS