Directability’s influence on human-agent trust
Y. Jiang (TU Delft - Electrical Engineering, Mathematics and Computer Science)
M.L. Tielman – Mentor (TU Delft - Interactive Intelligence)
Ruben S. Verhagen – Graduation committee member (TU Delft - Interactive Intelligence)
Carolina Centeio Jorge – Graduation committee member (TU Delft - Interactive Intelligence)
Jesse Krijthe – Coach (TU Delft - Pattern Recognition and Bioinformatics)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Mutual human-agent trust is of great importance for humans and agents to complete a task collaboratively. This paper aims at studying one of the factors influencing this mutual trust, the directability of humans to agents. Previous studies either take directability as a general concept without looking
into its different representations or fail to bridge the gap between directability and trust. This paper starts by analyzing directability’s different representations including commands, suggestions, and warnings, then investigates their influences on trust respectively. The experiment is set up in Block World For Teams(BW4T). Afterward, the trust is measured both by calculating the risk-taking behaviours by humans and using questionnaires. The result from the experiment suggests that collaborating with directability improves trust from humans to agents. Among different directability representations, commands and suggestions are the best ways to boost trust. However, the confounding factors
such as familiarity with the experiment also make a difference to the final result, those factors can be further investigated in the future.