Ronen Grinberg, Dan Waisman, Rinat B. Rosenberg-Kima E67 המחברים רוצים לציין באופן מיוחד את רונן גרינברג ז"ל שעבד באופן מסור על הנושא הנחקר במסגרת הדוקטורט שלו על הכשרה לשיח אמפתי בעזרת AI של סטודנטים לרפואה בטכניון ובמסגרת עמותת כוונים. רונן היה איש חינוך אהוב ודגול, בעבודתו ניהל את מחלקת מחקר, פיתוח והדרכה בעמותת כוונים, המכשירה צעירים עם מגבלה לחיים עצמאיים בקהילה, היה אדם כותב ויצירתי, איש משפחה וחברים . רונן האמין ביכולתה של הבינה לעזור לבני האדם להיות אנושיים יותר. רונן נהרג בתאונת דרכים ב - .18/11/25 Introduction In modern rehabilitation settings, educational rehabilitation counselors play a critical role in fostering the personal, emotional, and social development of individuals with disabilities going through complex transitions (Hartley & Tarvydas, 2022). Because of their direct interpersonal support for vulnerable individuals, the ability to communicate with empathy becomes a crucial professional skill rather than simply a desirable personal trait. Advances in educational technology, especially the integration of AI and conversational agents, have created new opportunities for communication training in social professions. AI-powered chatbots have been successfully used to support learning across various fields, offering learners safe, repetitive practice environments with immediate, personalized feedback (Liaw et al., 2023). These systems can simulate emotionally detailed interactions, making them particularly effective for fostering empathy in professionals (Levkovich et al., 2025; Pira et al., 2025). Vygotsky’s Sociocultural Theory (Vygotsky, 1978) suggests that higher cognitive functions, such as professional communication, develop through social interaction with a More Knowledgeable Other (MKO). But can a digital agent serve as a social entity? the Media Equation Theory posits that people automatically and unconsciously treat computers, media, and artificial entities as if they were real social beings (Reeves & Nass, 1996) and the Computers Are Social Actors (CASA) framework (Nass & Moon, 2000) further explains that users apply social norms, expectations, and interpersonal scripts to computers. Rooted in this theoretical framework, we introduce a chatbot-based system designed to train educational rehabilitation counselors in empathetic communication. This research aims to evaluate the quality and limitations of the VP as perceived by eight senior social workers. The following question guided this qualitative evaluation: How do social workers evaluate the credibility and appropriateness of the Virtual Participant (VP)? Method Participants Eight senior social workers from Kivunim (https://www.kvn.org.il/), a nonprofit organization that runs preparatory programs for independent living, participated in this study. The social workers were the system's main users and had already worked with it in practice. The participants signed a consent form approved by the university ethics committee. Materials The study used the Cesura.ai platform (https://cesura.ai/). The system's conversational and clinical reasoning capabilities were powered by Google's Gemini 2.5 Pro large language model. Two senior social workers, one specializing in cerebral palsy (CP) and the other in autism spectrum disorder (ASD), answered the agent-builder's prompts and supplied example dialogue to define the personas for two distinct VPs: a young adult with CP and a young adult with ASD. During the interaction, the
RkJQdWJsaXNoZXIy Mjk0MjAwOQ==