Feb. 24, 2023 — What does artificial intelligence (AI) ​“think” about its impact on the next-generation workforce? If you ask ChatGPT, the buzzy new chatbot developed by OpenAI, it spouts out a few examples of how AI could disrupt the job market and then closes with some advice:

“Overall, it is difficult to predict exactly how AI will impact students’ career paths, but it is likely that it will bring both challenges and opportunities. It is important for students to stay informed about developments in AI and to consider the role that AI could play in their future careers.”

Finding and embracing new opportunities to leverage AI in the science realm is one of the motivations behind the ​“Intro to AI-Driven Science on Supercomputers” training series hosted by the U.S. Department of Energy’s (DOE) Argonne National Laboratory. Aimed at undergraduate and graduate students across the country, the series is designed to introduce a new generation to using AI and high performance computing (HPC) technologies for science.

“At Argonne, we have these world-class computing systems and experts who are enabling researchers to do some very cool science. And these resources are available to anyone in the world who has a research problem that requires powerful HPC or AI systems,” said Paige Kinsley, educational outreach lead at the Argonne Leadership Computing Facility (ALCF), a DOE Office of Science user facility. ​“The training series gives us a chance to share these amazing capabilities with a new crowd while also teaching them how to use such tools in their own research.”

Fast-developing AI technologies, like the large language models that power ChatGPT, are opening the door to new possibilities in scientific research and beyond. For example, an Argonne-led team recently developed the first genome-scale large language models to explore the evolutionary dynamics of COVID-19. Their groundbreaking work was one of the use cases presented at the Argonne training series late last year, giving attendees a look at a real-world project that is using AI and supercomputers to advance science for the benefit of society.

“The biggest takeaway I had is the progress scientists are making using supercomputers and neural networks,” said Zachary Wooten, a Ph.D. student studying statistics at Rice University, who attended the AI training program. ​“Many times in the series I was blown away by guest lecturers explaining how they use the supercomputers at Argonne for their research. I sincerely believe the breakthroughs that are being made through supercomputers are going to solve so many problems that society will look very different in the next couple of decades.”

Comprised of eight virtual classes, the 2022 program welcomed over 200 attendees from 90 universities for hands-on instruction using methods, such as AI training and neural networks, on ALCF supercomputers.

For Wooten, the series presented an opportunity to learn how HPC could help advance his team’s efforts to create an automated tool for planning radiation treatments, which is typically a costly and time-consuming process. Developed in collaboration with researchers at the University of Texas MD Anderson Cancer Center, the tool is meant to help radiation clinics generate therapy treatment plans for patients in low- and middle-income areas.

“Our tool works through the use of various deep learning algorithms. Essentially, we input medical images of the patient and the algorithm outputs the appropriate radiation treatment plan,” Wooten said. ​“Naturally, understanding how to better use deep learning algorithms on a supercomputer at Argonne would only help to better understand how to use the computer cluster at MD Anderson.”

While the series is targeted at students, it is open to anyone interested in gaining experience using AI for science. Kishor Kapale, department chair and professor of physics at Western Illinois University, signed up to learn more about how machine learning can be applied to various research problems. In his case, he has been working with collaborators from the California Institute of Technology and Google to develop an AI system that can play quantum chess.

“I needed to understand the machine learning paradigms in a bit more detail at a level more manageable to me due to time constraints,” Kapale said. ​“I really enjoyed the fast-paced hands-on training sessions. They kept me on my toes.”

In addition to bolstering his AI skills, Kapale got an opportunity to learn how ALCF supercomputers could benefit his research moving forward.

“Getting acquainted with the ALCF’s HPC systems was an added bonus,” he said. ​“This training offered me an extensive introduction to these machines and how to use them to get things done.”

For those who were unable to attend the 2022 program, the ALCF has posted videos from all of the sessions on YouTube and materials for the hands-on activities on GitHub. Stay tuned to the ALCF training page for details on the 2023 Intro to AI-Driven Science on Supercomputers, which will be offered in the fall.

About Argonne

The Argonne Leadership Computing Facility provides supercomputing capabilities to the scientific and engineering community to advance fundamental discovery and understanding in a broad range of disciplines. Supported by the U.S. Department of Energy’s (DOE’s) Office of Science, Advanced Scientific Computing Research (ASCR) program, the ALCF is one of two DOE Leadership Computing Facilities in the nation dedicated to open science.

Argonne National Laboratory seeks solutions to pressing national problems in science and technology. The nation’s first national laboratory, Argonne conducts leading-edge basic and applied scientific research in virtually every scientific discipline. Argonne researchers work closely with researchers from hundreds of companies, universities, and federal, state and municipal agencies to help them solve their specific problems, advance America’s scientific leadership and prepare the nation for a better future. With employees from more than 60 nations, Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.

---

Source: Jim Collins, Argonne National Laboratory