Round Table Discussion

Training the next generation of scientists in the era of Generative & Agentic AI

17th Workshop of the
France-China Particle Physics Network and Laboratory

2 July 2026, IP2I

Panelists: Julien DONINI and Fazhi QI
Animator: Eric Kajfasz and Julien Zoubian
Slides link

Context

From Alexandre Boucaud's JDEV presentation

"Fully autonomous AI science produces… noise (for now)" [...]
"but with a human in the loop, the hints are already striking" [...]
"AI is changing fast — don't bet on "now"

https://aboucaud.github.io/slides/2026/lightcone-research-jdev/
JDEV 2026 — Lightcone Research

Goal for today discussion

  • Focus on education, not technology
  • Share perspectives from research, supervision and students
  • Leave with a few concrete actions in the framework of the FCPPN/L

How the discussion will work

  1. 3 discussion topics
    ~5 min each with the panelists
  2. Open discussion
    Students, supervisors, researcher, engineers, everyone is welcome to contribute
  3. Wrap-up
    Let's identify a few concrete actions for training the next generation of scientists.
The goal is not to answer every question, but to start a conversation and leave with actionable ideas.

1. Learning to code in the age of LLMs

Should students still learn programming from scratch when LLMs can generate code?

5 minutes

2. Writing papers and theses with AI

Is it acceptable for a PhD student to use an LLM to write parts of a paper or thesis?

5 minutes

3. Agentic AI and future scientific skills

If autonomous AI agents become part of daily research, which human skills become more important?

5 minutes

Open Discussion

  • Has AI already changed the way you work?
  • What skills do students struggle with today?
  • What training would be most useful?
  • What should institutions do differently?

~10 minutes

Take away actions

https://mensuel.framapad.org/p/h6bwed9kag-am50

Shared Notes QR code

~5 minutes