AI is rapidly becoming a critical layer of national infrastructure, yet the mechanisms by which we make AI systems accountable remain immature and fragmented. As frontier AI models reshape research, education, science, and industry, new national initiatives such as the National AI Research Resource (NAIRR) and the Genesis Mission are expanding access to AI capabilities while raising urgent questions about accountability, transparency, and public trust.

Drawing on five years as a co-PI of the NSF Intelligent Cyberinfrastructure with Computational Learning in the Environment (ICICLE) AI Institute, Plale will discuss how AI infrastructure is evolving across the edge, cloud, and high-performance computing continuum. ICICLE has explored not only the technical challenges of composing AI functionality across distributed environments, but also the governance challenges that emerge when models, data, and workflows move across organizational and computational boundaries.

Of particular interest is AI accountability: accountability requires that AI systems carry a memory of their origins, assumptions, limitations, and prior uses. The talk reflects on the US research and educational landscape, touching on several questions:

• What infrastructure is needed for trustworthy AI?
• How should universities position themselves in an era dominated by frontier models?
• What role should higher education play in balancing innovation with accountability?

The next generation of AI infrastructure must support not only scale and capability, but also transparency, accountability, and public trust. Universities have a unique opportunity to help build that future.

Beth Plale is Executive Director of the School of Computer and Data Sciences at the University of Oregon and a nationally recognized leader in computational and data science, cyberinfrastructure, open science, and trustworthy AI. Before joining Oregon, she held the Michael A. and Laurie Burns McRobbie Bicentennial Chair in Computer Engineering at Indiana University Bloomington, where she chaired the Department of Intelligent Systems Engineering and served as Executive Director of the Pervasive Technology Institute, a leading university research organization supporting advanced computing and data-enabled discovery. Her research advances the computational and data infrastructure underpinning AI, high-performance computing, open science, reproducibility, and AI accountability. She has held leadership roles with the NSF ICICLE AI Institute, the Research Data Alliance, and the HathiTrust Research Center. From 2017 to 2022, she served at the U.S. National Science Foundation as Science Advisor for Public Access. Her honors include the U.S. Department of Energy Early Career Award; she is a Senior Member of ACM and IEEE.

The NSF-funded READINESS (Revolutionizing AI-Driven Autonomous Experimentation for Next-Generation Semiconductor Synthesis) Programmable Cloud Laboratory (PCL) at Rice University aims to transform materials synthesis from traditional trial-and-error approaches into intelligent, data-driven, autonomous experimentation. By integrating advanced robotics, automated synthesis and characterization, digital twins, shared data infrastructure, and self-improving AI agents, READINESS is developing a cloud-accessible platform for the discovery and scalable manufacturing of next-generation electronic and quantum materials, including 2D semiconductors, oxides, and diamond thin films.

A central component of this effort is a robust research computing and data ecosystem that connects physical laboratories with AI models, digital twins, and autonomous decision-making. Through standardized, remotely programmable access to experimental facilities, READINESS seeks to democratize advanced materials research, accelerate scientific discovery, and bridge the gap between laboratory innovation and semiconductor manufacturing.

This presentation will introduce the READINESS vision, its planned capabilities and early priorities, and emerging opportunities for interdisciplinary collaboration in autonomous scientific discovery.

Jun Lou is the Karl F. Hasselmann Professor of Materials Science and NanoEngineering at Rice University and director of the Nanomaterials, Nanomechanics and Nanodevices Laboratory (N³L). His research spans the synthesis and design of 2D and other nanomaterials; nanomechanical and multiphysics characterization; advanced materials and devices for biomedical, electronics, energy, quantum, and structural applications; and AI for manufacturing. He has published more than 350 peer-reviewed papers, with over 68,000 citations and an h-index of 120. He directs the NSF Programmable Cloud Laboratory (PCL) project READINESS—Revolutionizing AI-Driven Autonomous Experimentation for Next-Generation Semiconductor Synthesis—which integrates AI, robotics, digital twins, and shared data infrastructure. He is also Editor-in-Chief of Materials Today.

Reproducibility in AI requires attention to the software, data, and procedures that connect a model to a research result. Recording a model name or preserving a conversation provides only part of the information needed to understand, repeat, and assess an AI-assisted analysis. This talk examines reproducibility across the research workflow, connecting software stacks and data management to the evaluation of AI outputs.

Dr. Christopher S. Simmons will discuss how to document computational environments, identify model and configuration versions, preserve data provenance, and distinguish repeatable execution from reproducible conclusions. The discussion will also consider appropriate handling of sensitive research data, explicit checks on generated outputs, and the role of human judgment when results vary.

Attendees will leave with a practical framework for deciding what to record, what to test, and what to share so that collaborators can examine and repeat AI-assisted work. The emphasis is on practices researchers can apply within their available computing environments.

Dr. Christopher S. Simmons is Cambridge Computer's Scientist-in-Residence, where he works with research institutions on high-performance computing, AI, and research data infrastructure. He earned his Ph.D. in Physical Chemistry at the University of Texas at Austin and served as research faculty and Deputy Director of the Center for Predictive Engineering and Computational Science at UT Austin's Oden Institute. He later led research computing at UT Dallas before joining MIT as Deputy Executive Director of the Massachusetts Green High Performance Computing Center. A founder and former project lead of OpenHPC, he remains active as its outreach and engagement coordinator and volunteers with the Open Storage Network.