Student Experiences of an AI Mediated PatientBot Simulation for Interviewing Skills in Psychology and Social Work Education A South African Pilot Study
This South African pilot study found that while an AI-mediated patient chatbot simulation was generally perceived as usable and beneficial for building interviewing confidence and reflective skills among psychology and social work students, its effectiveness is currently limited by interactional rigidity and typing burdens, suggesting it works best as a supplementary tool alongside clear orientation and traditional face-to-face practice.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the training of future psychologists and social workers, a critical skill is the ability to sit with a person in distress, ask the right questions, and listen with genuine empathy. These professionals must learn to build trust, navigate complex emotions, and guide a conversation without causing harm. Traditionally, this learning happens through role-playing with classmates or supervised practice with real clients, but these opportunities are often scarce, high-pressure, and difficult to repeat. To bridge this gap, educators are increasingly turning to artificial intelligence. This technology can create a simulated environment where students practice their interviewing skills against a computer program that acts as a patient. The goal is not to replace human interaction, but to provide a safe, low-risk space for students to make mistakes, receive immediate feedback, and refine their approach before they ever meet a real person in need.
A recent pilot study at a university in South Africa explored how students experienced such a tool, known as a "PatientBot." The researchers wanted to understand if this digital simulation could effectively support the training of psychology and social work students. They focused on a specific group of eleven students, mostly women, who used the chatbot to practice interviewing. The students were given a simulated client with a specific background and were asked to conduct an interview, asking questions and responding to the client's emotional cues. The chatbot was designed to respond in the language the student chose, allowing for practice in English or one of the student's home languages, a crucial feature in a country with twelve official languages. After the session, the bot provided feedback on the student's performance, pointing out missed opportunities or suggesting better ways to phrase questions.
The students reported that the experience was generally positive and useful. They found the system easy to use and appreciated the psychological safety of practicing with a machine rather than a peer or a lecturer. One student noted that the bot helped them identify areas for improvement that they might have missed otherwise. The feedback was particularly valued because it highlighted specific gaps in their questioning, such as failing to explore a topic deeply enough or closing a line of inquiry too quickly. This immediate, personalized feedback allowed students to recognize their own blind spots and adjust their strategies in a low-stakes environment. Many participants felt that the simulation boosted their confidence, making them feel less anxious when they later practiced with real people or in role-playing exercises.
However, the study also revealed significant limitations that prevent the bot from being a complete substitute for human interaction. The primary challenge was the text-based nature of the conversation. Students reported that typing their questions took a long time and that the lack of voice, facial expressions, and body language made the interaction feel rigid and less realistic. In fields like social work and psychology, reading non-verbal cues is essential, and a text-only interface cannot replicate the nuance of a person's tone, hesitation, or posture. While the bot was helpful for practicing the structure of an interview and the wording of questions, it could not teach students how to respond to the emotional atmosphere of a room.
The researchers concluded that while the PatientBot shows promise as a supplementary tool, it works best when integrated into a broader training program rather than used in isolation. The study suggests that the most effective approach is a blended model: students use the chatbot for initial, repetitive practice to build confidence and learn the basics of questioning, followed by structured reflection and feedback from a human instructor. This human oversight is vital to ensure the feedback is accurate and to help students interpret the limitations of the simulation. The study also highlighted the need for more diverse scenarios that reflect the complex realities of different cultures and languages, ensuring that the training prepares students for the wide variety of people they will eventually serve.
Ultimately, the findings indicate that artificial intelligence can serve as a valuable stepping stone in professional education, offering a safe space for students to rehearse and refine their skills. It is not a magic solution that solves all training challenges, nor does it replace the need for human connection. Instead, it acts as a supportive scaffold, allowing students to make mistakes and learn from them without the fear of hurting a real client. As the technology evolves, incorporating voice interaction and more realistic emotional responses could further enhance its value, but for now, its greatest strength lies in its ability to provide a structured, repeatable, and judgment-free environment for learning the delicate art of human connection.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.