Is this project an undergraduate, graduate, or faculty project?
Graduate
Project Type
individual
Campus
Daytona Beach
Authors' Class Standing
Ulker Yusupova, Graduate student
Lead Presenter's Name
Ulker Yusupova
Lead Presenter's College
DB College of Arts and Sciences
Faculty Mentor Name
Dr. Briana Sobel
Abstract
This study explores how people perceive Artificial Intelligence (AI) therapists compared to human therapists, focusing on differences in empathy, effectiveness, and competence. As AI tools like ChatGPT and mental health chatbots become more commonly used for emotional support, it is important to understand how users evaluate them in comparison to human professionals. To begin addressing this, in the first part of the study, participants reviewed therapist-client conversation transcripts and identified whether they believed the therapist was human or AI-generated, helping us validate the materials for the main study. Notably, initial findings suggest that participants frequently relied more on surface-level language cues such as repetition, structure, and phrasing rather than deeper therapeutic qualities when identifying AI responses. In contrast, human therapists were more often associated with a more natural conversation flow and open-ended questioning. Building on these results, the first part has now been completed, and the collected data is being used to design the second part, where participants will rate and compare their perceptions of AI and human therapists more directly. Ultimately, these findings aim to provide a better understanding of the role AI can play in mental health support, including both its potential and its limitations.
Did this research project receive funding support (Spark, SURF, Research Abroad, Student Internal Grants, Collaborative, Climbing, or Ignite Grants) from the Office of Undergraduate Research?
No
Artificial Intelligence (AI) in Mental Health: Perceptions of AI therapists
This study explores how people perceive Artificial Intelligence (AI) therapists compared to human therapists, focusing on differences in empathy, effectiveness, and competence. As AI tools like ChatGPT and mental health chatbots become more commonly used for emotional support, it is important to understand how users evaluate them in comparison to human professionals. To begin addressing this, in the first part of the study, participants reviewed therapist-client conversation transcripts and identified whether they believed the therapist was human or AI-generated, helping us validate the materials for the main study. Notably, initial findings suggest that participants frequently relied more on surface-level language cues such as repetition, structure, and phrasing rather than deeper therapeutic qualities when identifying AI responses. In contrast, human therapists were more often associated with a more natural conversation flow and open-ended questioning. Building on these results, the first part has now been completed, and the collected data is being used to design the second part, where participants will rate and compare their perceptions of AI and human therapists more directly. Ultimately, these findings aim to provide a better understanding of the role AI can play in mental health support, including both its potential and its limitations.