During my summer residency as part of the AI program at CSU Channel Islands, I discovered playlab.ai, a platform that enables users to create their own chatbot AIs. An initial example demonstrated a chatbot designed to answer syllabus-related questions for students—a useful tool indeed. However, I immediately saw the potential for a more innovative application: creating virtual patients for students to engage with.
This fall of 2024, I am teaching a Disability Services course that traditionally uses case studies from my clinical experiences as a social worker. The course encompasses a wide range of topics, including disability diagnoses, philosophies, policies, practices, interviewing skills, and supportive resources. While role-playing can sometimes feel awkward for students, it effectively brings theoretical concepts to life.
Recently, I utilized playlab.ai to create “Michael AI,” a chatbot patient modeled after a real client I had facing a significant life transition—moving from his aging parents’ home to a group home. This past week marked the inaugural session where students interacted with Michael AI. Tasked with interviewing him and developing an Individual Support Plan, students worked in groups, strategizing their approach to the conversation. Unlike traditional role-plays, the interaction with Michael AI evoked genuine emotional engagement from the students. They were not only impressed by the technology but also prompted to explore realistic professional scenarios. For instance, when Michael AI mentioned that only his parents would have certain answers to the students’ questions, a student inquired about the appropriate steps if a real patient provided a similar response. This led to a valuable discussion about involving family members in subsequent meetings.
Feedback from this session highlighted areas for improvement: Michael AI was initially programmed with limited goals, indicating a need for richer background information to enhance the realism of future interactions. Additionally, some responses were too structured, resembling bullet points and headings typical of AI outputs. I addressed this by adjusting the AI’s settings to promote more natural, conversational interactions.
For the students’ upcoming midterm assignment, I’ve developed a scenario involving another AI-generated couple, Santiago and Sofia Veranda, allowing students to practice their skills with both a patient and his spouse. They will conduct interviews, complete a Biopsychosocial Assessment, and prepare an Individual Support Plan for Santiago. I am eager to see how these enhancements will further enrich our students’ learning experience.
As the new AI-focused Digital Learning Mentor at CSUCI, I am excited to assist other educators in leveraging playlab.ai and similar technologies to amplify the impact of their teaching. Email me any time at kristen.linton@csuci.edu.
