Answers are grounded in this paper and cite their sources.
As AI–human interactions (AI–HI) increase, little is known about how they may shape people’s own empathic responding, particularly in counselor education. This exploratory pilot study examined counselor-facilitated encounters with AI-generated personas. Ten adults, four of whom were graduate counseling students, each completed one facilitated session with two text-based persona encounters. Empathy was self-assessed before and after the session with eight items from the Perspective Taking and Empathic Concern subscales of the Interpersonal Reactivity Index. Within each encounter, an embedded Unlock Tier System (UTS) scored participants’ responses on eight empathy signals that governed how much the persona disclosed, and two researchers coded the transcripts. Mean scores rose from 23.10 (SD = 5.02) to 25.80 (SD = 2.90), a mean change of 2.70 points (95% CI [0.54, 4.86]; Cohen’s dz = 0.89). Nine of ten participants reached the deepest UTS tier, and transcript coding showed frequent Perspective Taking and Empathic Concern. Because the study had no comparison group and measured outcomes immediately after a single session, these results describe short-term, in-session empathic responding rather than lasting change. The UTS is offered as a preliminary framework that requires further validation.
No takes yet. Share an insight, caveat, or question.
Luster et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: