As artificial intelligence (AI) continues to permeate many aspects of our everyday lives, building an AI-ready workforce has become a national priority.

To help grow the community of researchers who can leverage AI for science, the U.S Department of Energy’s (DOE) Argonne National Laboratory offers its annual Intro to AI-Driven Science on Supercomputers training series. Since its inception in 2021, the series has hosted over 600 attendees for an immersive learning experience focused on the fundamentals of using AI and supercomputers. Organized by the Argonne Leadership Computing Facility (ALCF), a DOE Office of Science user facility, the series is aimed at undergraduate and graduate students enrolled at U.S. universities and community colleges.

“Drawing on Argonne’s extensive AI expertise and computing capabilities, we designed our AI training series to equip a new generation of researchers with the knowledge and skills needed to harness the power of AI and high performance computing (HPC) for scientific discoveries and innovations,” said Paige Kinsley, educational outreach lead at the ALCF.

The 2024 series, which concluded this spring, was comprised of eight virtual sessions that gave attendees hands-on experience running AI models on ALCF supercomputers and AI accelerators. Led by Argonne experts, the sessions covered topics such as large language models, neural networks, training AI models and novel AI accelerators, including the systems housed at the ALCF AI Testbed. Each session was followed by a science-focused talk that highlighted how different Argonne teams are using AI for research in fields such as biosciences, climate and cosmology. To extend the program’s reach, videos of each session are available on YouTube.

Over 250 participants attended the 2024 series with many receiving digital certificates for completing the program’s homework assignments. Below, we highlight some of the attendees’ experiences and how the program will impact their careers moving forward. 

John Wu

With dreams of pursuing a career in precision medicine, John Wu, a student at Blinn College in Texas, is immersing himself in the realm of AI to understand its potential for optimizing medical treatments and improving patient outcomes.

“I believe AI will aid physicians in evaluating and refining treatment regimens, significantly enhancing survivability rates,” he said.

At Blinn College, Wu is studying mathematics to complement his background in the biosciences. When he heard about the ALCF AI Training Series, he jumped at the opportunity to learn more about the current state of AI for science.

“I was drawn to the series because it is unique in that it offered a broad overview of the AI landscape, but also invaluable practical experience in training and using AI models,” he said. “Without a comprehensive understanding of these AI models, how can one identify the most suitable model to fine-tune or train for the benefit of patients?”

In the field of precision medicine, Wu noted the need to analyze large and complex datasets, such as medical records and genetic information, to identify potential health issues and develop personalized treatment plans. Due to the vast amounts of data involved, the research can require powerful HPC systems. He valued the opportunity to gain hands-on experience with ALCF supercomputers during the training series. 

“Training these AI models on a personal laptop is often not feasible due to the lack of computing power and necessary hardware requirements,” Wu said. “It is rare, especially for students, to be given access to state-of-the-art supercomputers, such as Argonne’s Polaris, to train and tune these models.”

Akshata Tiwari

Akshata Tiwari has a lengthy history of taking part in Argonne’s educational opportunities. It began in high school when she attended the lab’s weeklong Big Data Camp. She then returned to Argonne in the summer of 2023 as a University of Illinois Urbana-Champaign (UIUC) student participating in DOE’s Science Undergraduate Laboratory Internship (SULI) program.

The internship led to her current role as a part-time research aide in Argonne’s Energy Systems and Infrastructure Analysis division, where she is working with a team to use AI to analyze thousands of battery patents.

“My research is focused on using machine learning techniques to classify patents into one of 13 different battery chemistries,” Tiwari said. “Our goal is to gain insights into the leading battery technologies being developed around the world and inform new research that will help keep the U.S. at the forefront of the green energy revolution.”

At UIUC, she is studying computer science and economics, with an eye toward a future career that leverages HPC and AI to advance research in the energy sector. Tiwari attended the ALCF AI training series to learn how advances in AI could help her current and future research endeavors.

“I was amazed at how these seemingly complex machine learning techniques can really be broken down to be very understandable for everyone,” Tiwari said. “I was able to understand it with much with more depth and clarity than I would have through online research or web tutorials.”

While Tiwari is embracing AI as a tool to enhance her research, she also noted the importance of acknowledging its limitations.

“At the end of the day, the science that's being done — from the initial design to final execution — still has to be done by the researchers themselves,” she said. “The creativity that comes from the human mind, the collaboration that arises, all the nuances involved in our thinking processes — are some of the things that will be very difficult for AI to replicate.”

Trung Vo 

For Trung Vo, the AI training series presented an opportunity to learn how AI could benefit his research involving molecular dynamics simulations. As a Ph.D. student at the University of Illinois Chicago, he is using these simulations to study how molecules behave at the interface between two liquids.

“Molecular dynamics simulations are a tool for scientists to explore and investigate real-life processes at the atomic level,” Vo said. “AI has opened up a new bright future for molecular dynamics and may be able to help overcome some of its limitations by enhancing its accuracy and reducing the amount of computing power needed for simulations.”

The training series not only gave him new skills and ideas to incorporate into his research but also broadened his perspective on the impact of AI on science.

“The sessions helped me bring me up to speed on the newest AI tools and technologies,” Vo said. “And the science talks were awesome. The program piqued my curiosity in many areas, expanded my understanding of applied AI in science and enhanced my overall skills as a scientist.”

Vo also took advantage of the networking opportunities presented by the series. This summer, he will be joining Argonne as a research aide to work on developing a large language model-based bot to help ALCF users with a variety of scientific computing tasks.

Alexandra Day

Alexandra Day’s Ph.D. studies at Northwestern University are focused on using machine learning to process large scientific datasets. Her research involves collaborating with university scientists to explore how AI can speed up the analysis of nanoparticle images. 

“I believe that AI has tremendous potential to find patterns in large, complex datasets,” Day said. “Using AI, we can automate and accelerate tedious tasks that currently require extensive human input.” 

She attended the ALCF training series to gain more experience with modern HPC systems and increase her knowledge about emerging AI tools and practices. 

“Since AI and HPC are at the core of my work, this training series was a perfect fit for my interests,” Day said. “I’ve already used some of what I learned in the training series in my research and referred back to the videos online.”

In addition, the series offered a valuable refresher on core concepts that will be useful as Day assists her advisor in teaching two AI courses at Northwestern. She also appreciated the diverse science talks that capped off each weekly session. 

“Since I’m beginning to explore post-Ph.D. career options, the training series gave me a great overview of how AI is being used in different scientific fields, from biology to physics,” Day said.