Ivan Habernal on the Edge of AI, Law and Music

Описание: What happens when a legal NLP researcher — who also happens to be a jazz bassist — steps into a conversation about law, AI, and creativity?

In this episode of Mondai Baigel, Ivan Habernal joins Michał Jackowski to talk about legal argument mining, privacy-preserving AI, and whether generative models are really as powerful as we think — especially in the courtroom.

Bonus: we explore what jazz improvisation teaches us about human agency, and why real-time risk-taking matters — in both law and music.

Takeaways
Legal NLP isn’t about replacing lawyers — it’s about helping us ask better empirical questions.
Privacy in AI isn’t solved — even anonymized court data can leak more than we think.
GenAI models underperform on specialized legal tasks — older, smaller models sometimes do better.
Improvisation in jazz = agency in judgment — what makes something meaningful isn’t just structure, but timing, intention, and risk.
We still need legal scholarship to guide how we build and evaluate AI tools in justice systems.

Chapters:
00:00 Introduction to AI, Law, and Music
01:16 Ivan's Journey into Legal NLP
02:59 Interconnection of Legal NLP and Privacy
09:40 Empirical Research on ECHR Judgments
13:34 Comparative Analysis of Czech Supreme Court Decisions
16:57 The Role of NLP in the Age of GenAI
23:39 AI's Impact on Music Creation
30:44 The Future of AI in Law and Music