UA Little Rock Researcher Studies How People Build, Lose, and Regain Trust in AI

UA Little Rock faculty member Dr. Bahareh Jozranjbar
After completing her year as the University of Arkansas at Little Rock's inaugural AI scholar-in-residence, Dr. Bahareh Jozranjbar has joined the Department of Psychological Science as an assistant professor, allowing her to build on the research and AI literacy initiatives she developed during the residency. Photo by Benjamin Krain

From asking a chatbot for advice to using AI to help with a school assignment, people increasingly face the same question: When should they trust the technology, and when should they question it?

At the University of Arkansas at Little Rock, Dr. Bahareh Jozranjbar is studying what happens when people put their trust in AI — and helping students learn to recognize when that trust is justified.

After completing her year as UA Little Rock’s inaugural AI scholar-in-residence, Jozranjbar has joined the Department of Psychological Science as an assistant professor, allowing her to build on the research and AI literacy initiatives she developed during the residency.

“The scholar-in-residence appointment was designed as a one-year inaugural position, and it has formally concluded,” Jozranjbar said. “I’m delighted to say that its mission hasn’t concluded at all — it has grown.”

Supported by the department’s Marie Wilson Howells Endowment, the residency gave Jozranjbar a year to investigate how people assess AI-generated information and decide when to rely on the technology.

During her residency, she developed and piloted an AI literacy curriculum in introductory psychology courses, established a research program focused on trust in AI systems, and contributed to university-wide conversations about the technology’s role in higher education.

“The residency was about focused exploration; the faculty role is about building something lasting — a research lab, courses, and a pipeline of students who can do this work themselves,” Jozranjbar said.

One of the central questions driving Jozranjbar’s research is how people determine whether to trust information generated by AI.

While Jozranjbar initially expected the greatest challenge to be people relying too heavily on AI, her research has also drawn attention to the opposite problem: people losing trust in the technology after encountering a mistake, even when the system may be reliable in other situations.

Both responses can affect how effectively people use the technology.

Jozranjabar has also observed that students often accept confident-sounding AI responses without checking them against independent sources, even though the technology’s confidence doesn’t necessarily reflect its accuracy.

“The goal of AI literacy isn’t teaching people to trust AI more or less, it’s teaching them to trust it accurately — to match their reliance to the system’s actual competence in a given moment,” she said.

Jozranjbar is applying that approach in two courses this semester: Applied Psychology, and Statistics and Methods I.

Rather than allowing students to immediately turn to AI for answers, she asks them to develop their own conclusions first, consult an AI tool, and then explain whether the information changed their thinking.

“In my courses, students commit to their own answer or argument before they’re allowed to consult an AI tool, and then they have to articulate why they did or didn’t revise,” Jozranjbar said. “That does two things: it protects the independent thinking that learning requires, and it makes their reliance on AI visible to them, which is the first step in calibrating it.”

Her work in the classroom also builds on the AI literacy curriculum piloted during her residency.

Jozranjbar said feedback and data from the pilot are still being evaluated, with the next steps focused on refining the materials and determining how they might fit into broader university efforts to support AI literacy.

Beyond the classroom, she is continuing to investigate how trust in AI develops, changes, and recovers after a system makes a mistake.

One of her current research projects examines how people respond to AI-generated advice while making quick visual decisions. By tracking participants’ choices over time, she hopes to better understand how quickly people lose confidence in an AI system after an error, how long it takes to rebuild that trust, and whether a person’s own ability to complete the task influences their response.

“We all know the feeling — a GPS sends you the wrong way once, and suddenly you don’t quite believe it anymore, even when it’s right,” Jozranjbar said.

The study is moving into data collection, and Jozranjbar plans to eventually expand the research into a laboratory setting using technologies such as eye tracking, electroencephalography (EEG), and functional near-infrared spectroscopy (fNIRS). These tools could help researchers examine where people direct their attention and how brain activity changes as trust in AI is lost and rebuilt.

She is also collaborating on a National Science Foundation proposal led by a colleague at the University of Arkansas at Pine Bluff. The proposed project would examine whether psychology students at UAPB can accurately evaluate scientific explanations without knowing whether they were written by a person or an AI system.

Students would assess the explanations for accuracy, supporting evidence, and potentially overstated claims, with their evaluations compared against those of experts.

For Jozranjbar, becoming a faculty member also creates new opportunities to involve undergraduate students in that research, allowing them to gain hands-on experience studying the relationship between technology and human behavior.

Dr. Elisabeth Sherwin, chair of the Department of Psychological Science, said Jozranjbar’s appointment will expand opportunities for students to participate in research and explore new topics in the classroom.

“We are thrilled to have Dr. Jozranjbar join us as a full-time faculty member,” Sherwin said. “This brings her skills, knowledge, and expertise to our students and the department. Her research will provide our students another opportunity to participate in research, a goal we have been focusing on for a while.”

The department also plans to offer a dual-listed special topics course, Mind, Brain and Machine, taught by Jozranjbar in fall 2027.

Looking ahead, she hopes her work will help establish UA Little Rock as a leader in human-centered AI research while preparing students to use the technology thoughtfully in their education and future careers.

“AI is best understood as a fallible advisor, not an oracle and not a fraud,” Jozranjbar said. “These systems are genuinely useful and genuinely wrong some of the time, and the confidence in their tone tells you almost nothing about which is which.”

She hopes students will leave her courses with more than an understanding of how to use AI tools. She wants them to recognize the importance of their own judgment and the opportunities they have to help shape the technology’s future.

“Students shouldn’t just see themselves as users of AI,” she said. “Every field, including psychology, needs people who study, evaluate, and shape these systems. That’s a career path, and it’s open to them right now.”