Workshop report highlights a new way to co-design for scientific computing
More than 40 experts from national laboratories, universities, industry and community organizations met in a workshop last year to begin planning toward next-generation ecosystems for scientific computing.
The workshop was organized by Lois Curfman McInnes and collaborators as part of her 2024 DOE Office of Science Distinguished Scientist Fellowship. McInnes, a senior computational scientist and Argonne Distinguished Fellow in the Mathematics and Computer Science division at the U.S. Department of Energy’s Argonne National Laboratory, emphasized that she wanted to use the fellowship to explore how communities can collaborate in developing and managing reusable scientific software and applications.
“The scientific software community is being challenged by rapid developments such as heterogeneous computing architectures and emerging AI technologies,” McInnes said. “Addressing these challenges successfully requires a team effort of the entire workforce — scientists, developers and users.”
The workshop findings are now available in a report that captures the community requirements and recommendations for accelerating scientific software development.
A New Kind of Co-Design
Participants proposed a forward-looking approach called socio-technical co-design. Traditional co-design in high-performance computing (HPC) focuses mainly on improving the technology — hardware, software, algorithms and applications — by having scientists, developers and stakeholders work closely together. Socio-technical co-design expands this idea by including social and organizational factors as part of the process. It recognizes that progress depends not only on technology but also on collaboration, communication and workforce development. This means that technical innovation and human factors — such as team dynamics and training — are developed together, not separately.
Building on traditional approaches, this broader approach interweaves technical and social elements throughout all aspects of work, while closely coupling cycles of R&D innovation between computing technologies and driving applications.
AI’s Role in the Future of Scientific Computing
A major theme in the report is how artificial intelligence (AI) is transforming science — and what’s still missing to fully benefit from it. Current challenges include lack of standardized interfaces for data and software integration, gaps between how humans and AI systems communicate, and limited efforts to teach scientists how to create and use AI-generated code effectively.
To create a strong and adaptable scientific computing ecosystem for the next decade, the workshop participants outlined several identified priorities:
-
Near-term (1–2 years): Launch pilot projects on hybrid AI/HPC software infrastructure as well as cross-disciplinary collaboration and pedagogy for AI-driven scientific computing, establish responsible AI guidelines and prototype public–private partnerships.
-
Mid-term (3–5 years): Explore scaling approaches for modular software ecosystems for AI/HPC, and introduce workforce training programs for researchers and students.
-
Long-term (5+ years): Develop community-wide frameworks for AI governance and explore ways to make AI agents true research partners.
A Call for Bold Thinking
The report ends with a challenge to the scientific community: Embrace risk-tolerant, creative approaches that combine human ingenuity with AI innovation. “By intentionally interweaving technical and social components in next-generation scientific computing, we can create feedback loops that accelerate both scientific discovery and real-world impact,” the report says.
McInnes and her colleagues envision two more workshops over the next two years, with the goal of mapping pathways toward robust, cross-disciplinary software ecosystems.
“Building on insights from the 2025 workshop report, we will evaluate strategies for creating scalable software ecosystems,” McInnes said. “We will follow a co-design methodology, integrating topics in team-based scientific software, AI in scientific computing and community/workforce development. Our aim is to achieve a truly holistic approach.”
For more information, see the “Report of the 2025 Workshop on Next-Generation Ecosystems for Scientific Computing: Harnessing Community, Software, and AI for Cross-Disciplinary Team Science,” L. C. McInnes et al., 2025, https://doi.org/10.48550/arXiv.2510.03413.
Argonne National Laboratory seeks solutions to pressing national problems in science and technology by conducting leading-edge basic and applied research in virtually every scientific discipline. Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.
The U.S. Department of Energy’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science.