Artificial intelligence is expected to play a defining role in the future of fundamental physics research and in unlocking some of the universe’s deepest mysteries, according to the incoming director-general of CERN.
British physicist Professor Mark Thomson, who took over the leadership of CERN on January 1, 2026, says that machine learning is opening up entirely new possibilities in particle physics, developments he compares to the AI-driven breakthroughs in protein structure prediction that earned scientists at Google DeepMind the Nobel Prize in Chemistry last October.
At the Large Hadron Collider (LHC), these same techniques are already being used to detect extraordinarily rare phenomena considered crucial for understanding how particles acquired mass after the Big Bang, as well as for exploring theoretical scenarios about the long-term stability of the universe itself.
A new accelerator on the horizon
Thomson’s remarks come as CERN’s council advances plans for the Future Circular Collider — a proposed new accelerator stretching 90 kilometers in circumference, far surpassing the LHC. Despite skepticism from some quarters, following what critics describe as a lack of dramatic discoveries since the identification of the Higgs boson in 2012, Thomson believes AI is breathing new life into the search for physics beyond the Standard Model.
He anticipates that major breakthroughs could come after 2030, when a planned upgrade to the LHC will increase the intensity of its particle beam tenfold. This will enable unprecedented observations of the Higgs boson — known popularly as the “God particle” — the field-defining particle that grants mass to all other particles and holds the fabric of the universe together.
The God particle meets itself
“There is a specific measurement of the Higgs boson that is fundamental to the nature of the universe,” Thomson explained. “We will attempt to produce not one, but two Higgs bosons simultaneously.”
This would allow scientists to measure for the first time how the Higgs particle gives mass to itself, a phenomenon known as Higgs self-coupling. While the simultaneous production of two Higgs bosons is exceedingly rare, Thomson said he is now confident it is achievable: “Five years ago I would have considered this beyond the reach of the LHC. Now I am certain we will obtain a reliable measurement.”
The strength of this self-coupling is critical to understanding how, a trillionth of a second after the Big Bang, particles first acquired mass. It could also reveal whether the Higgs field is in a stable state or whether it might undergo a new transition in the future, a scenario that would result in the annihilation of the universe as we know it. The Standard Model indicates this is theoretically possible, though there is no cause for immediate concern.
“It is not something that could happen on any timescale relevant even to our stars,” said CERN theoretical physicist Dr. Matthew McCullough. “But it remains a purely scientific question: could it happen?”
Thomson described it as “a deeply fundamental property of the universe that we have not yet fully understood,” adding that if the measurement of the self-coupling deviates from theory, it would constitute “a monumental discovery.”
AI at every stage
Artificial intelligence is now being applied at every stage of LHC operations — from selecting which data to collect, to analyzing it afterward. “When the LHC collides protons, it produces around 40 million collisions per second, and we have to decide within a microsecond which events are worth keeping,” explained Dr. Katharine Leney from the ATLAS experiment. “Thanks to AI, we have advanced at least 20 years beyond what we originally expected to achieve.”
Scientists also hope the LHC may one day produce dark matter, which is believed to make up a large portion of the universe. Because its nature remains unknown, the search is inherently difficult. Generative AI could help by enabling more open-ended research approaches. As Thomson put it: “Instead of searching for a specific signal, we can ask: is there anything unexpected in this data?”
