Study compares human and AI-simulated clients in counselling role-plays

Researchers compared recorded counselling training conversations involving human role-players and a large language model, looking at both client responses and novice counsellors’ communication.

What the study found

In an arXiv preprint, Eric Rudolph, Philipp Steigerwald and Jens Albrecht report that counselling communication patterns showed both similarities and differences across role-plays with human and AI-simulated clients. The abstract does not specify which patterns differed, in what direction, or by how much.

The researchers first recorded novice counsellors interacting with peers acting as clients, then recorded role-plays involving clients simulated by Mistral’s Mixtral 8x7b model using 4-bit quantization. They analysed the interactions sentence by sentence using a counselling communication pattern system. Their questions concerned how closely model responses reproduced conversational dynamics and whether counsellors communicated differently with the two kinds of client.

Why it matters

Role-playing gives counsellors a way to practise conversations. The authors suggest that language-model simulations have potential as training exercises and that their comparison can help identify areas for refining virtual clients. The abstract does not establish that the exercises improved counselling competence.

Limits and unknowns

The abstract does not report the number of participants or sessions, the specific communication patterns measured, numerical results, or how the analysis was validated. Human peers playing clients are also a role-play comparison, not evidence of how either scenario matches conversations with actual clients. The abstract does not say whether the observed differences would matter for training outcomes.

Read the research

Authors: Eric Rudolph, Philipp Steigerwald, Jens Albrecht

Published: 2026-10-29

Source basis: Based on the abstract only

Original abstract

This study investigates the capabilities of Large Language Models to simulate counselling clients in educational role-plays in comparison to human role-players. Initially, we recorded role-playing sessions, where novice counsellors interacted with human peers acting as clients, followed by role-plays between humans and clients simulated by Mistrals Mixtral 8x7b using 4-bit quantization. These interactions were analysed with a counselling communication pattern system at sentence level. We investigated two key questions: (1) to what extent LLM-generated responses replicate authentic conversational dynamics and (2) whether counsellors' communication behaviour differs when interacting with human versus LLM-simulated clients. The findings highlight both similarities and differences in the application of counselling patterns across scenarios, showing the potential of LLM-based role-playing exercises to enhance counselling competencies and to identify areas for further refinement in virtual client simulations.

This article is an AI-generated adaptation, not a substitute for the original research. Preprints may not have been peer reviewed; AI can make mistakes.