Study compares human and Mixtral-simulated clients in counselling role-play
Original title: Comparing Human Role-Players and LLM-Simulated Clients in Online Counselling Training: An Analysis of Counselling Patterns
Novice counsellors practiced with peers and with 4-bit Mixtral 8x7b clients; sentence-level patterns both matched and diverged.
Researchers recorded novice counsellors role-playing with human peers, then with clients simulated by Mistral's Mixtral 8x7b using 4-bit quantization. Interactions were coded with a counselling communication pattern system at sentence level. The study asks how far the model reproduces conversational dynamics and whether counsellors change behaviour with a simulated client. It reports similarities and differences and treats the setup as a training aid that still needs refinement.
- Clients were simulated with Mixtral 8x7b at 4-bit quantization.
- Analysis used a counselling pattern system at sentence level.
- One question is whether counsellor behaviour changes with an LLM client.
- Findings note both similar and different counselling-pattern use.
- The work is framed as educational role-play, not clinical deployment.
- Vendors
- Mistral
- AI models
- Mixtral 8x7b
From the original article
1 min read at arxiv.orgThis 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.
By Eric Rudolph · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.