On the last day of the inaugural AI4Fusion Summer School at William & Mary, Jim Slone ’26 felt confident that commercial nuclear fusion energy could really be 30 years away this time.
Slone, a double major in physics and data science, was among the 14 undergraduates selected to attend the fully funded, two week intensive course that took place June 3-14, leveraging the role of artificial intelligence and machine learning in controlling the variables involved in fusion experiments.
Fusion technology has the potential to provide safe and reliable clean energy on Earth: Fusing hydrogen nuclei into helium is how the sun releases energy. In her lecture notes, W&M physicist Saskia Mordijck referred to the “sun in the jar” as the focus of the summer school.
Coordinated by the W&M data science program and held at Miller Hall, the school is an integral component of a $5 million project funded by the Department of Energy and involving Mordijck, Data Science Assistant Professor Cristiano Fanelli and Computer Science Associate Professor Pieter Peers. Featuring internal and external experts in data science and nuclear fusion, the summer school will return to William & Mary for at least two more summers; interested applicants for next year’s iteration are encouraged to visit the school’s landing page in the coming months.
Mordijck, Class of 1955 Associate Professor of Physics at William & Mary and also the president of the University Fusion Association, lauded the “automatic synergy” between data and fusion scientists at the summer school.
She aims to replicate this interdisciplinarity in the William & Mary proposed School of Computing, Data Sciences, and Physics.
“Let’s make these units better by being together,” she said. “I’ve had the most fun conversations about potential research collaborations that I did not expect to have.”
Sometimes, these stemmed from talking about a simple dataset. At the summer, school, real datasets were made accessible for students to work on, as the school alternated lectures and hands-on work based on real data. As part of its Vision 2026 strategic plan, William & Mary is focused on promoting data fluency as a key skill across disciplines.
“Certain topics were used pedagogically for students, but at the same time they may also open possibilities for research,” said Fanelli, who in addition to organizing the school served as lecturer. This activity was part of his growing research portfolio, including two more projects on which he serves as PI and co-PI, aiming to accelerate experimental nuclear physics with machine learning.
At a recent White House summit, Mordijck had highlighted knowledge exchange as a key factor in getting to the goal of commercial nuclear fusion. Data access and interoperability are a real focus of the DoE project, which is committed to making datasets findable, accessible, interoperable and reusable.
The project’s focus on accessibility also meant providing summer school students with appropriate access to computer resources. Fanelli and his data science partners arranged for Jupyter notebooks for interactive programming to be used on a Kubernetes cluster that was made available to class attendees. To ensure the broader impact of this activity, a Jupyter book was set, containing all lectures and notebooks, and can be accessed online. Also, to take action on participation barriers, the school organizers had also reached out to The Computational Research Access Network.
The first edition of the summer school was attended by an array of students from William & Mary and other eight institutions across the United States. For W&M students like Michael Campagna ’26, a physics and mathematics major, the course reaffirmed an interest in nuclear fusion while Rebecca George ’26, a computer science and mathematical biology major, seized the opportunity to take data science classes while learning more about nuclear fusion.
“An introduction to these topics has been really important because there are questions we don’t have answers to yet and this is how you start,” said Nathan Cummings, an external collaborator and lecturer from the United Kingdom Atomic Energy Authority. He defined nuclear fusion and artificial intelligence as the fields of research with the biggest potential impact on human technological advancement.
“It’s just evident to me that in the future, machine learning is going to be an important tool for pretty much everyone to use,” said Campagna.
Antonella Di Marzio, Senior Research Writer