Paper Type
Complete
Abstract
As agents are increasingly embedded in user journeys, reducing overreliance on potentially incorrect algorithmic output without harming user experience is essential. This article proposes extending the user journey to include verification of algorithmic output. We argue that uncertainty cues (e.g., “I’m not sure…”) motivate users to verify responses, thereby changing the user journey and experience. An online experiment (N = 196) and a field study (N = 145) on verification behavior reveal positive effects on user experience but show that users who verify outputs through agents’ explanations report lower trust and reliance than those who check information sources. The opposite is true for perceptions of journey seamlessness, pointing towards a trade-off between trust and seamlessness perceptions when selecting how to verify. We highlight the role of user expertise in reducing overreliance and enabling efficient verification, emphasizing the importance of developing basic expertise for meaningful, reliable interactions with agents.
Paper Number
1549
Recommended Citation
Asisof, Alina; Sou, Davinny; and Nißen, Marcia, "Keeping Agents Honest: The Role of Uncertainty Cues for AI Interactions" (2026). AMCIS 2026 Proceedings. 12.
https://aisel.aisnet.org/amcis2026/sig_hci/sig_hci/12
Keeping Agents Honest: The Role of Uncertainty Cues for AI Interactions
As agents are increasingly embedded in user journeys, reducing overreliance on potentially incorrect algorithmic output without harming user experience is essential. This article proposes extending the user journey to include verification of algorithmic output. We argue that uncertainty cues (e.g., “I’m not sure…”) motivate users to verify responses, thereby changing the user journey and experience. An online experiment (N = 196) and a field study (N = 145) on verification behavior reveal positive effects on user experience but show that users who verify outputs through agents’ explanations report lower trust and reliance than those who check information sources. The opposite is true for perceptions of journey seamlessness, pointing towards a trade-off between trust and seamlessness perceptions when selecting how to verify. We highlight the role of user expertise in reducing overreliance and enabling efficient verification, emphasizing the importance of developing basic expertise for meaningful, reliable interactions with agents.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG HCI