E68 Credibility and Appropriateness in AI-Based Simulation as Empathy Training Tools (Short paper) VP delivers its verbal responses as text, accompanied by a description of its non-verbal behaviors (e.g., position, tone of voice, eye movements) enclosed in brackets. For example: "Fine, as usual, nothing special happened. [Lifts her head slightly, glances for a moment toward the counselor, and then returns to her phone]". Data Collection and Analysis Data were collected in a single one-hour focus group with eight senior social workers from Kivunim, seven of them had gained hands-on experience with the tool prior to the session. The focus group followed a multi-stage process. The social workers received one of two transcripts of interactions between a VP and a practitioner and were asked to assess individually the VP’s credibility. A fullgroup discussion on the credibility and appropriateness of both VPs followed. The group discussion was recorded and transcribed verbatim for analysis, and the transcripts annotated with the social workers' individual notes were also collected. The data were analyzed using thematic analysis (Braun & Clarke, 2006). The process included close reading of the transcripts, identifying and grouping initial codes, and developing themes that reflected recurring insights. Findings Five themes emerged from the thematic analysis with respect to the research question ("How do social workers evaluate the credibility of the Virtual Participant (VP)?") (see Table 1): (1) Verbal credibility - Some VPs' responses were judged by the social workers as incongruent with expected linguistic and cognitive profiles, described as "too complicated" or overly articulated, while others were perceived as vague or stereotypical, (2) Emotional and behavioral credibility - Emotional and behavioral cues were also experienced by the social workers as exaggerated or artificial, (3) Credibility in representing resistance and initiative – The participants noted that the portrayal of resistance often appeared linguistically rigid rather than behaviorally grounded, (4) Inconsistent verbal utterances and described behavior - A further source of difficulty was the inconsistency between the VP’s verbal utterances and the accompanying behavioral descriptions, (5) Absence of physical presence - A recurring theme was the absence of physical presence, which they described as a central aspect of their work with young adults. In their professional practice, subtle embodied cues provide crucial information about a participant’s emotional state, and (6) Absence of shared history - The social workers also emphasized the difficulties posed by the VP's lack of personal history. In practice, they build a deep familiarity with each participant’s personal history, routines, and idiosyncrasies, knowledge that grounds their interpretations and guides empathetic responses. The VP, however, could not "remember" previous encounters, refer to earlier conversations, or reflect a personalized daily schedule. Without this continuity, the VP felt detached from the lived context of the preparatory programs and therefore less authentic as a relational partner.
RkJQdWJsaXNoZXIy Mjk0MjAwOQ==