Project Highlights
Our research explores AI technologies with a focus on both their theoretical underpinnings and practical applications. Each project reflects our commitment to responsible AI, ensuring these technologies serve the broader public good. Below are highlights of our current and recent projects.
AI in Mental Health Screening
In collaboration with healthcare providers, this project examines whether AI-powered chatbots can make mental health screening faster and more scalable. Users often perceive AI agents as more judgmental than human counselors, even when the responses are identical, which lowers disclosure and adherence to recommended treatment. The findings point to the need for empathetic, carefully designed AI in sensitive clinical settings.
Contact: Alan Dennis (ardennis@iu.edu)
Celebrity Digital Humans in Customer Service
This project tests whether giving AI agents a celebrity likeness, such as a digital Hugh Jackman, changes how customers respond. Users were more willing to engage with celebrity-inspired digital humans, and service errors hurt satisfaction less. The results matter for retail and entertainment brands, while raising ethical questions about manipulation and the line between reality and simulation.
Contact: Alan Dennis (ardennis@iu.edu)
Reducing Polarization through Generative AI
These studies examine whether conversational agents powered by generative AI can reduce political polarization in discussions of controversial topics. The agents improved mutual understanding, fostered empathy, and reduced hostility. The work has implications for public policy, civic education, and social media moderation, and highlights the challenge of designing AI that is both neutral and effective on sensitive issues.
Contact: Antino Kim (antino@iu.edu)
AI in Teamwork and Productivity
This project studies AI as a collaborative team member in brainstorming and idea generation. In a pilot study, about 25% of the ideas teams ultimately selected originated from AI contributions. By summarizing key points and surfacing overlooked information, AI teammates can boost creativity and efficiency in collaborative work.
Researchers: Alan Dennis, Warren Rosengren, Tucker Todd, Joseph Steed
Contact: Alan Dennis (ardennis@iu.edu)
Tolerance for Error in Human-Like AI
This research examines how users react when human-like AI systems make mechanical mistakes, such as mishearing or mistyping. People are more forgiving of errors from human-like AI, but human-like cues also shape interactions in unexpected ways: users may express more frustration or aggression, and may feel pressure to conceal their knowledge or individuality.
Contact: Alfred Benedikt Brendel (alfbrend@iu.edu)
AI Errors and User Trust
This study shows that the timing of AI errors matters: early mistakes erode user trust far more than later ones, even when the system performs accurately afterward. Letting users adjust how much they rely on AI recommendations increases satisfaction and willingness to keep using the system, underscoring the importance of user autonomy.
Contact: Antino Kim (antino@iu.edu)
Articles
We pride ourselves on producing world-class research that not only advances academic knowledge but also provides actionable insights for businesses, policymakers, and technologists. Below are some of our featured articles.
Less Artificial, More Intelligent: Understanding Affinity, Trustworthiness, and Preference for Digital Humans
Information Systems Research
This study explores the deployment of highly realistic digital human agents (DHAs) controlled by artificial intelligence for customer service tasks. Four major experiments were conducted to examine user perceptions of trustworthiness, affinity, and willingness to work with DHAs. A mixed-method approach was used, combining surveys, interviews, observations, and neurophysiological measurements. The study tested four different DHAs, including commercial products and future-oriented designs. Findings highlight the importance of transparency and design maturity in fostering user trust and engagement.
AI Agents as Team Members: Effects on Satisfaction, Conflict, Trustworthiness, and Willingness to Work With
Journal of Management Information Systems
This research investigates how team members perceive AI agents as virtual teammates for managing information, coordinating processes, and performing tasks. The study uses a 2x2 lab experiment to analyze the effects of team member type (human vs. AI), performance levels, and interactions with other team members. Results show that AI team members were seen as having higher ability and integrity but lower benevolence, with no significant differences in trustworthiness or willingness to work with them. However, AI team members led to lower process satisfaction, though they reduced conflict when performing well compared to human teammates.
Understanding Affinity, Trustworthiness, and Preference for Digital Humans
Information Systems Research
This paper, authored by Mike Seymour and collaborators at Iowa State, University of Sydney, and Indiana University, explores how AI-driven digital human agents are perceived and responded to by users in online interactions. Through a series of experiments, including EEG brain analysis and surveys, the researchers found that users respond positively to digital humans, especially when they closely resemble and behave like real people. The study highlights that people often don't mind if these digital agents are AI-driven, as long as they act in a human-like manner. The paper discusses implications for developing friendly, engaging digital humans that are increasingly preferred over traditional chatbot.