The Deep Underground Neutrino Experiment (DUNE) at Fermi National Accelerator Laboratory has begun installing the structural elements of its massive neutrino detector modules while integrating artificial intelligence into measurements, rapid detection, and operations. The AI tools are meant to help scientists process the high-resolution data DUNE will produce and reach the experiment's discovery goals faster.
Neutrinos interact only through the weak force and gravity, which makes them hard to detect. Trillions pass through the human body every second, but only one might interact in a lifetime. DUNE uses intense sources, such as Fermilab's accelerator complex, and massive detectors to increase the odds of capturing them.
The experiment consists of a near detector at Fermilab and a far detector installed a mile underground at the Sanford Underground Research Facility in South Dakota. Both use liquid-argon time projection chamber technology to record the rare interactions between neutrinos and argon atoms.
Tracking particle interactions
Liquid-argon TPCs capture neutrino interactions well, but reconstructing the signals is complicated. DUNE will produce images with high resolution, which adds detail but also makes reconstruction harder.
"We are going to have very high resolution, almost photographic-quality images of these interactions," said Leigh Whitehead, DUNE AI/ML co-lead. "Which is great, but also brings challenges such as making reconstruction more difficult because we see such fine detail."
AI algorithms will recognize when a neutrino interaction takes place inside the detector, analyze the resulting particle tracks, trace back to where the interaction occurred, and extrapolate the energy and direction of the neutrino. These tools are orders of magnitude faster than traditional methods, improving signal processing, event classification, and particle identification. The approach builds on earlier Fermilab experiments, including NOvA and MicroBooNE, which were among the first high-energy physics experiments to use a deep neural net to identify particle interactions. Researchers working on AI for Science & Research can see how these methods carry over from those experiments to DUNE's larger scale.
"With DUNE, researchers are taking innovative approaches to try and impact every aspect of the experiment," said Sowjanya Gollapinni, co-spokesperson for DUNE. "These approaches are going to really accelerate DUNE in terms of the commissioning process and the sensitivity of the experiment to reach our full discovery potential."
Locating exploding stars
Beyond reconstructing common neutrino signals, AI is being trained to identify rare events. These triggers are fast decision-making systems that sort through thousands of interactions and save only the most interesting for physics.
One major trigger would fire if a supernova goes off in our galaxy. When a large star runs out of fuel, it explodes and sends stellar remains into space. Photons get trapped in the resulting gas and dust, but neutrinos pass through and arrive at Earth a few hours before the light burst. The early alert lets astrophysicists point their telescopes in time to see the light.
An AI algorithm will constantly read the detector, searching for interesting interactions. The instant it suspects a supernova signal, it tells the system to record data from 10 seconds before and 100 seconds after the candidate signal. The neutrinos may also provide insight into the stellar object left behind, either a neutron star or black hole.
Running the detectors
DUNE's detectors are made of thousands of components that must be monitored during data collection. If an error occurs, operators need to respond quickly. The collaboration is exploring whether a large language model can scan logbooks of documented fixes and provide citations to solutions. Fermilab is coordinating with six other national labs to develop a similar tool for particle accelerators.
The collaboration is also studying whether machine learning can predict detector anomalies before they happen. The Department of Energy's Genesis Mission is supporting these efforts across national labs and universities.
"The Genesis Mission has strengthened the already existing collaboration across the national labs and universities on AI efforts," said Gollapinni. "It is taking everything to the next level and will maximize our capabilities to accelerate DUNE."
DUNE is also training the next generation of researchers. The collaboration includes more than a thousand people from around the globe and trains hundreds of students each year. For research scientists looking to build similar skills, the AI Learning Path for Research Scientists covers data modeling, lab automation, and experimental design.
Why this matters for Science and Research
DUNE shows how AI can be applied across an entire experimental program: real-time triggering, event reconstruction, anomaly prediction, and operational support. The methods being developed here - deep neural nets for particle identification, LLMs for logbook search, and predictive machine learning for detector health - are directly transferable to other large-scale research facilities. Scientists working in data-heavy fields can expect similar AI integration to become standard in experimental design and operations.
Your membership also unlocks:
