The CIRSS speaker series will continue in Spring 2026 with the ongoing theme of “The AI Disruption.” Our speakers will discuss how recent advances in AI have reshaped their research — what has been made easier and what has become more difficult — and reflect upon its broader disruptive impact on society.
We meet most Wednesdays, 9am-10am US Central Time, on Zoom. This event is open to the public, and everyone is welcome to attend. The series is hosted by the Center for Informatics Research in Science and Scholarship (CIRSS) at the School of Information Sciences at the University of Illinois at Urbana-Champaign, and led by Yuanxi Fu and Timothy McPhillips . If you have any questions, please contact Timothy McPhillips.
Participate: To join a live talk, follow the “Join Here” link for the current week below to access the iSchool event page for the talk. From there, click the “PARTICIPATE online” button to join the live Zoom session. Recordings of past talks can be found via the “Recording” links below if available.
Follow: To receive weekly updates on upcoming talks, subscribe to our CIRSS Seminars mailing list at https://lists.ischool.illinois.edu/lists/info/cirss-seminars. Subscribe to add events to your calendar via Google Calendar or Outlook.
Spring 2026 Speakers
Sarah B. Lawsky, University of Illinois Urbana-Champaign
Wednesday January 28th 2026, 9am-10am CT
Title: Computational Law, Transparency, and Accountability: The Case of Direct File
Abstract: What role can, or should, tools based on large language models play in the administration and formalization of law? To begin to wrestle with that question, this talk will look at a project that uses no such tools at all: Direct File. Direct File was a program created by the United States government that allowed some taxpayers to file their federal income tax returns online, for free. In 2025, the Internal Revenue Service publicly posted almost the entire computer code underlying the Direct File program. The Direct File code shows how electronic filing can, contrary to common wisdom, increase transparency. The computer code itself, if released publicly, creates transparency by revealing behind-the-scenes choices that are merely implicit in forms, instructions, and other informal guidance. And some computer code for electronic filing, such as the code for Direct File, allows visualization and explanatory tools to be built on top of that code, making the application of the law and various administrative choices more transparent even to those who are not comfortable reading computer code. The existence, transparency, and legibility of Direct File code and its outputs poses a challenge to large language models in the area of law administration; the presentation is thus intended to begin a conversation among those present about in what ways, or whether, the administration of law should undergo an “AI disruption.”
Bio: Sarah B. Lawsky, the L.B. Lall and Sumitra Devi Lall Professor of Law at the University of Illinois College of Law, studies tax law, computational law, and the intersection of the two. Her recent work focuses on the formalization of tax law. Professor Lawsky’s research arguing for using a particular nonstandard logic to formalize tax law is the conceptual foundation for the domain-specific programming language Catala, which is the project of a team of computer scientists and lawyers. Before joining the University of Illinois, Professor Lawsky taught at Northwestern Pritzker School of Law, UC Irvine School of Law, and George Washington University Law School. Before entering academia, she worked as a tax lawyer for large law firms. For more information, visit the personal website of Professor Lawsky: https://www.sarahlawsky.org/
Yue Guo, University of Illinois Urbana-Champaign
Wednesday February 4th 2026, 9am-10am CT
Title: Fluent but Not Understandable: the Limits of LLMs in Health Communication
Abstract: Health information is widely available, yet it often fails to support understanding and decision-making by non-experts. In this talk, I show that LLM-generated health explanations— while often fluent and even preferred by readers— do not reliably improve human comprehension compared to expert-written summaries. Through controlled evaluations and large-scale human studies, I identify a key reason for this gap: current automated evaluation and optimization methods favor surface quality over human-centered criteria such as understanding and actionability. Moving beyond diagnosis, I present LLM-based personalization as a path forward, demonstrating how integrating external knowledge and lightweight model adaptation to readers’ backgrounds can better support meaningful understanding and actionable health decisions. Together, this work shows that aligning LLM objectives, evaluation, and personalization is critical for effective health communication.
Bio: Yue Guo, MBBS, PhD, is a physician-scientist and Assistant Professor in the School of Information Sciences. Her research lies at the intersection of clinical medicine, health informatics, and natural language processing, with an emphasis on developing computational methods that make medical information more accessible, trustworthy, and actionable for clinicians, patients, and researchers. Dr. Guo earned her PhD in Health Informatics from the University of Washington, following her medical training (MBBS) at Capital Medical University and epidemiology training at Johns Hopkins University. Before joining the University of Illinois Urbana-Champaign, she conducted research at Microsoft Research, Google, and the Allen Institute for AI (AI2). Her scholarly contributions have been recognized through multiple honors, including the AAAI New Faculty Highlight, Edward H. Shortliffe Doctoral Dissertation Award finalist, and the Arnold O. Beckman Research Award.
Nataliya Kosmyna, Massachusetts Institute of Technology
Wednesday February 11th 2026, 9am-10am CT
Title: GenAI: Friend or Foe to Your Brain
Abstract: For millions of years, human intelligence set the standard. But now, the lightning pace of tech has left us gasping, struggling to keep up with our own cognitive demands. AI has pushed civilization into overdrive, yet what we are ultimately doing is burning terawatts of power on data centers and excluding humans from this growth. We have built systems that are prefixed ‘smart’, but not smart enough to break free from their own inefficiency.
In this talk, Dr. Nataliya Kosmyna will argue that we need to start creating more seamless AI interfacing directly with our brains, achieving the same outcomes with the brain’s energy consumption levels.
Technology should amplify our creativity, not snuff it out. It should fuel social interactions, not isolate us. The goal is not to replace human thought, but to propel us into a Type II civilization.
Instead, we are trapped in a dystopian remix of 1984 — 2025’s version — where digital censorship and surveillance threaten to choke innovation in nations that refuse to play along.
This talk will explore critical questions: What should define ownership in the age of AI and at what cost?
It is time to reclaim the conversation — because true evolution should never be about creating more artificial intelligence. It is about evolving the most powerful source of intelligence: Your Mind.
Bio: Dr. Kosmyna is a Research Scientist at MIT Media Lab’s Fluid Interfaces group and a Visiting Faculty Researcher at Google. She has over 15 years of experience in developing and designing end-to-end brain-computer interfaces (BCIs). Coming from a background in artificial intelligence, neuroscience and human-computer interaction (HCI), she is passionate about the idea of creating a partnership between AI and human intelligence, a fusion of the machine with the human brain.
Nataliya obtained her Ph.D in 2015 in the domain of non-invasive Brain-Computer Interfaces (BCIs). Most of her projects are focused around BCIs in the context of consumer grade applications. Nataliya is a public speaker, author of multiple research papers and reviewer of numerous professional journals and conferences. Dr. Kosmyna often collaborates with teams from Boston Dynamics, Microsoft Research, NASA. Additional information about Nataliya’s work is available on her MIT people page.
Jonathan Chen, Stanford University
Wednesday February 25th 2026, 11am-12pm CT
Title: AI in Medicine – Integrated Intelligence or Illusory Imitations?
Abstract: Pandora’s box has opened in the form of publicly available generative AI systems for every imaginable (and many unintended) purposes. With a global scarcity of medical expertise against the unlimited demand of people in need, AI’s potential to democratize healthcare knowledge, access, and to recover efficiencies is desperately needed. The implications are vast as we converge upon a point in history where human vs. computer generated content can no longer be reliably distinguished. This session will review the attention and intention required for AI applications in the high-stakes world of healthcare as we distinguish real magic from convincing illusions.
Bio: Jonathan H. Chen MD, PhD leads a clinical informatics research group to empower individuals with the collective experience of the many, combining human and artificial intelligence to deliver better care than either. Dr. Chen founded a company to translate his Computer Science graduate work into AI systems used by people around the world. His expertise is featured in the popular press with over 100 research publications and awards. Dr. Chen continues to practice medicine for the concrete rewards of caring for real people and to inspire his work to advance not just the science and technology, but to redefine how we train (and retrain) an entire generation of the healthcare workforce to safely and effectively integrate intelligent systems to reach all patients in need.
Rosina Weber, Drexel University
Wednesday March 4th 2026, 9am-10am CT
Title: XAI is in Trouble, but are LLMs?
Abstract: Researchers who study how artificial intelligence (AI) methods explain their decisions often discuss the field’s controversies and limitations. Some even contend that many publications offer little or no substantive contribution. In this talk, I illustrate the claim that explainable AI (XAI) is in trouble by describing four problems: disagreements over XAI’s scope, a lack of definitional cohesion, weak motivations for XAI research, and inconsistent evaluation practices. I then analyze the potential causes, linking these problems to the field’s interdisciplinary nature and to inadequate scientific rigor. This analysis yields a set of open research questions and recommendations for avoiding these problems. Finally, I discuss what this all means for the AI disruption we are now living.
Bio: Rosina Weber is a Professor of Information Science and Computer Science at Drexel University, where she advises students with interdisciplinary interests. Her interdisciplinary research spans legal, biomedical, and military domains, as well as academic and business problems. While investigating hybrid approaches, she has specialized in case-based reasoning (CBR) and explainable AI. Her research has been funded by DARPA, NIH, DHS, and international agencies. Her scholarship appears in venues such as AI Magazine, Applied AI Letters, Expert Systems with Applications, Knowledge-Based Systems, AAAI, and ICCBR. She has earned best-paper and research honors, co-chaired multiple XAI workshops, delivered XAI tutorials, and taught AI to both computer science majors and students in non-technical disciplines. She is an elected member of the AAAI Executive Council and a member of AAAS, AWIS, ACL, and NDM.
Florian Schnitzhofer, CEO Sysparency
Wednesday March 11th 2026, 11am-12pm CT
Title: The Vision of Self Driving Organizations
Abstract: In his talk, Florian Schnitzhofer reflects on his personal journey through the historical evolution of Artificial Intelligence, from early Bayesian email filter systems written in C to today’s emerging era of Agentic AI. Drawing on his experience as a management consultant, entrepreneur, investor, and researcher, he connects technological milestones with real world business and academic projects to illustrate how AI is reshaping the future of IT and software.
A central theme of the lecture is Agentic AI, meaning AI systems capable of autonomous decision making, goal oriented reasoning, and proactive action. Schnitzhofer explains how Agentic AI is already disrupting the traditional software development lifecycle by shifting from human written code toward AI driven system design, autonomous testing, self optimizing architectures, and continuous value delivery.
He introduces the concept of Self Driving Organizations, which are companies and public institutions that operate through AI powered digital twins, autonomous processes, and data driven governance structures. A key innovation presented in the lecture is the Digital Twin of Legislation, invented by Schnitzhofer. These digital legal twins translate laws into machine interpretable and executable systems, enabling automated administrative decisions and real time regulatory compliance. This approach has the potential to fundamentally transform governance, public administration, and the relationship between the state and society.
Ultimately, the talk explores the societal implications of Agentic AI and asks how autonomous systems will reshape economic structures, institutional design, and human roles. Schnitzhofer argues that we are entering a new era in which organizations and even states may become self driving, redefining value creation, management, and societal coordination in the age of intelligent machines.
Bio: Florian Schnitzhofer is a management consultant, serial entrepreneur, investor, and visionary. He is the owner of the ReqPOOL Group and Sysparency and invests in highly scalable software companies. He advises top management of leading organizations in Europe and US on AI strategy and digital transformation. As the inventor of the Digital Twin of Legislation and the visionary behind the concept of the Self Driving Company and State, he works at the intersection of artificial intelligence, governance, and organizational transformation.
Ha Nguyen, University of North Carolina at Chapel Hill
Wednesday March 25th 2026, 11am-12pm CT
Title: Whose Voice is the Chatbot? Designing AI for Science Learning
Abstract: Much discussion around generative AI (genAI) in education has focused on questions of adoption, such as how these tools should be used in classrooms. An equally important question is how AI systems can be designed to represent community perspectives and reflect the values that students and educators consider important for learning. In this talk, I will discuss a value-sensitive design approach to developing and evaluating genAI learning technology, the impact such technology has on student learning, and how AI interactions fit within the expanding set of digital tools students use. I will illustrate these ideas through a case study of designing chatbots to help high school students develop systems thinking and science communication within a climate change curriculum.
Bio: Ha Nguyen is an Assistant Professor of Learning Sciences & Psychological Studies in the School of Education at the University of North Carolina at Chapel Hill. Her research, most recently supported by the National Science Foundation, Spencer Foundation, and a UNC AI faculty fellowship, examines the design and impact of learning technologies, including AI and learning analytics, in STEM education, with particular focus on how technologies can support human-human and human-AI collaboration. Nguyen is Associate Editor of Behaviour & Information Technology and on the editorial board for the British Journal of Educational Technology. She received degrees from University of California-Irvine (PhD, MA) and Duke University (BA).
Hans-Georg Fill, University of Fribourg
Wednesday April 15th 2026, 9am-10am CT
Title: Lost in Generation: The Hidden Cost of AI and the Power of Conceptual Modeling
Abstract: Generative artificial intelligence has recently emerged as a global phenomenon. It has the capacity to generate texts, images, software, video, and audio content through statements in natural language without the need for technical expertise. However, these advancements come with costs, including increased energy consumption, potential copyright issues, or costs related to potentially false information. Additionally, there is a concern that humans are becoming overwhelmed by the sheer volume of information they are now generating individually, while at the same time experiencing a loss of skills and competencies that are assumingly being absorbed by AI. The field of conceptual modeling has a long tradition in eliciting and structuring knowledge in many domains for supporting human communication, for processing information, and reasoning about it. It is therefore well-suited for dealing with some of the challenges of generative AI. This includes support for AI input, better understanding of AI output, as well as preventing the deskilling of human actors. In this talk we will therefore discuss the past, present, and future of conceptual modeling in times of generative artificial intelligence and its contributions to information systems development.
Bio: Hans-Georg Fill is full professor for business informatics, vice-president of the Department of Informatics, and co-coordinator of the Smart Living Lab at the University of Fribourg, Switzerland. Before, he held positions at the University of Vienna, Austria and the University of Bamberg, Germany and was a visiting scholar at Stanford University, Karlsruhe Institute of Technology, Germany and the École des Mines at St. Etienne, France. He has received his PhD and habilitation from the University of Vienna, Austria. He has more than 15 years of experience in conceptual modeling and enterprise modeling both in academia and industrial research projects. He is supporting editor in chief of the gold open access journal Enterprise Modeling and Information Systems Architectures – International Journal of Conceptual Modeling (EMISAJ). He is speaker of the SIG Modeling of Enterprise Information Systems (MobIS) and deputy speaker of the Cross-sectional Technical Committee on Modeling (QFAM) of the German Informatics Society (GI).