AI Days 2026: large language models in mental health
At the AI Days 2026 of the University of Cologne, Joseph spoke about large language models (LLMs) in mental health.
Psychiatry is deeply language-driven: we diagnose, treat and understand patients through conversation, which makes LLMs particularly interesting for the field. The talk covered:
- language as a digital biomarker,
- how AI and clinicians make decisions,
- generative simulations of symptom trajectories.
The real challenge is not whether LLMs can do these things, but how we validate them rigorously and use them responsibly in clinical contexts.