In a recent article, Joachim Mnich, CERN’s Director of Research and Computing, states that artificial intelligence has become a strategic imperative at CERN. AI is now fundamentally integrated across scientific research, technical operations, and administrative functions. CERN has adopted a formal, organization-wide AI strategy to promote the responsible, open, and impactful use of AI. Other labs like Fermilab have similarly recognized AI as essential and established an organizational structure for AI.
Archi remains directly relevant in this context — it offers an open-source framework with a number of specific interfaces used in HEP and has been presented in joint meetings with CERN researchers. There is active interest in its use across several applications: supporting Computing Operations with onboarding new operators and assisting during standard operations, helping shift takers access critical information often buried in logbooks or operational instructions, and more.
One of the core lessons from this work is that building the framework is only the beginning — the real effort lies in tuning each specific application to be fast, reliable, and complete. Each deployment has a very specific purpose and needs to be carefully adapted.
There is significant activity now at the labs in AI applications and the project is on track to make an impact. Defining the application and tuning its performance — including benchmarking — will be the key components of the work ahead.
Originally published on the MIT PPC blog.