Participating Speakers
Dr. Shashank Priya
Speaker Bio:
Shashank Priya serves as the University of Minnesota’s vice president for research and innovation, overseeing a $1+ billion research enterprise across all campuses and facilities. He manages units responsible for administration of sponsored projects, research and regulatory compliance, technology commercialization, and corporate engagement, as well as 10 interdisciplinary academic centers and institutes. At UMN he has established the International Institute on Biosensing, the Biotechnology and Biomanufacturing Innovation Center, the Sustainable GeoCommunities program, a national security research initiative, a systemwide data science initiative, an artist-in-residence program, and a research partnership with North Carolina A&T, the nation’s largest historically Black university. Priya represents UMN research at a number of national and state organizations including APLU, AAU, the Council on Competitiveness, and MBOLD. He previously served as associate vice president for research and director of strategic initiatives, and professor of materials science and engineering at Penn State. He has published over 450 academic journal articles and holds 12 patents. He earned his BS from Allahabad University, his ME from the Indian Institute of Science, and his PhD in materials engineering from Penn State.
MN Secretary of State - Steve Simon
Talk Title:
AI and Elections
Talk Description:
Minnesota Secretary of State Steve Simon will discuss the role of the Office of the Secretary of State in administering Minnesota elections system, recent changes to Minnesota elections laws including those that relate to AI, and the real and potential impacts of artificial intelligence on Minnesota’s elections.
Speaker Bio:
Steve Simon is currently Minnesota's 22nd Secretary of State, inaugurated on January 5, 2015.
In his role as chief elections administrator, he collaborates across political affiliations and geographical boundaries to protect and enhance the right to vote. Simon oversees elections, business services, and the "Safe at Home" address confidentiality program. His goals include expanding voting access, removing barriers, streamlining business services, and enhancing protections for domestic violence victims.
Prior to his current position, Simon served a decade in the Minnesota House of Representatives, focusing on election issues and leading key reforms. His diverse career includes serving as Assistant Attorney General, private legal practice, and receiving multiple accolades for his advocacy.
Simon holds a B.A. in Political Science from Tufts University (1992) and a J.D. from the University of Minnesota Law School (1996). He resides in Hopkins with his family.
Mark Gardner
Talk Title:
Cracking the Code: AI, Medical Devices, and the Evolving FDA Landscape
Talk Description:
This presentation explores the growing role of artificial intelligence (AI) in medical devices and provides strategies for navigating the FDA's evolving regulatory framework for AI-powered medical technology. It offers valuable insights into the approval pathways for AI-based devices, including 510(k) clearance, De Novo classification, and premarket approval (PMA), as well as the FDA's guidance on predetermined change control plans (PCCP), software as a medical device (SaMD), and clinical decision support software (CDS).
Speaker Bio:
Mark has worked in healthcare for 25 years, including 10 years working in commercial roles at medical device and laboratory companies, and 15 years working in private legal practice. He advises FDA-regulated companies on regulatory, compliance, and privacy transactional and litigation matters. He founded Gardner Law and several operating and early-stage investment companies. Mark is an adjunct professor at three schools.
Nancy Sims
Talk Title:
Copyright & Ethics in Generative AI
Talk Description:
Permission, fair use, licensing, and book scanning are just a few of the issues involved in how source materials get -in- to content models. On the output side there are issues of copyrightability, originality, authorship, credit, citation, and much more. The training and tuning of content models also presents issues around equity, safety, and cross-border worker exploitation. In an interactive session, we’ll review some broadly applicable basics of copyright law, and briefly explore some considerations beyond academic ethics.
Speaker Bio:
Nancy Sims is the University of Minnesota Libraries' subject specialist on copyright issues. Nancy is both a librarian and a lawyer, with long experience working in academic libraries, and is fascinated by the pervasiveness of copyright issues in modern life. Nancy's role at UMN is to help individuals and groups understand how copyright affects their work, across all the breadth of creative outputs at a large University. At UMN and beyond, Nancy advocates for policies and practices that support sustainable scholarship, democratic information access, and wide public cultural participation.
Dr. Baobao Zhang
Talk Title:
Deliberative Democracy Can Increase Participation in the Governance of Artificial Intelligence: Evidence from a U.S. National Public Assembly
Talk Description:
We present the findings from the first-ever national public assembly on Artificial Intelligence (AI) governance in the U.S., focusing on categorizing risks from AI systems. While public opinion surveys are useful for understanding public attitudes toward AI, they are often limited in capturing surface-level preferences because many respondents lack technical knowledge about AI. Therefore, this study sought to explore public attitudes and policy preferences regarding high-risk AI through a public assembly. Through a nationally representative recruitment survey of 3,000 adults in the U.S., 40 individuals were randomly selected to participate. The participants learned from 8 experts (computer scientists and AI ethicists/policy experts) and deliberated under the guidance of facilitators, promoting a deeper understanding of the topic. The assembly participants indicate that general-purpose AI systems pose higher risks than narrow-purpose AI and that the government should determine accountability if an AI system harms people. Furthermore, a two-wave panel study found that the public assembly increased self-reported knowledge about AI, support for governmental regulation, and willingness to take political action to ensure safe AI development and deployment.
Speaker Bio:
Baobao Zhang, a Klarman Postdoctoral Fellow associated with the Department of Government, stands at the forefront of research in the governance of artificial intelligence (AI). In the upcoming Fall of 2021, she is set to embark on a new role as an assistant professor at the Maxwell School of Citizenship and Public Affairs at Syracuse University.Her current research is deeply entrenched in understanding the dynamics of AI governance, with a specific focus on public and elite opinions regarding AI. Baobao explores how the American welfare state can adapt to the evolving landscape of labor automation, showcasing a keen interest in the intersection of technology and societal structures. Baobao's scholarly contributions extend beyond AI governance, covering diverse topics such as the politics of the U.S. welfare state, attitudes towards climate change, and survey methodology. Her prolific work has found its way into prestigious publications, including PLOS One, Political Analysis, the Journal of Artificial Intelligence Research, and Nature Climate Change, reflecting the breadth and impact of her research endeavors.
Michael Corey
Talk Title:
The process is the product: How Mapping Prejudice uses community co-creation for reparative change
Talk Description:
Mapping Prejudice uses its Deed Machine software to sift through millions of historic property records to find and map racist property restrictions. While it seems like this would be an ideal candidate for AI-based automation, the project team has instead doubled down on its commitment to its community of more than 9,000 human volunteers. Technical Lead Michael Corey will discuss how the project's core mission, to build new awareness of structural racism that will lead to reparative change, has led the team to put human work at the center of its technical development. With demand from teams across the country for its technical support, Mapping Prejudice grapples every day with the challenge of how to scale the work of changing human minds.
Speaker Bio:
Michael Corey is the Geospatial, Technical and Data Lead / Associate Director for Mapping Prejudice. Michael's primary role is to design, build and maintain the Deed Machine, Mapping Prejudice's technical platform for identifying and mapping racial covenants, as well as to work with community partners and public records custodians to facilitate more racial covenants research. Before transitioning to public history, Michael spent 20 years as a journalist and data journalist at the Star Tribune, Reveal from the Center for Investigative Reporting and the Des Moines Register. His journalism work spanned zoning and segregation, mortgage disparities, the U.S.-Mexico border fence system, human-induced earthquakes, and sexual abuse in the Catholic Church.
Dr. Jisu Huh
Talk Title:
GenAI and Future of Advertising
Talk Description:
Dr. Huh's presentation discusses AI-driven transformations in the world of advertising and four general areas of advertising practice and research that GenAI is fundamentally altering: (1) consumers' experience of advertising; (2) societal and policy implications related to the truthfulness of AI-generated content; (3) the analytical considerations of data and algorithms; and (4) the functioning of the advertising industry and ethical issues.
Speaker Bio:
Dr. Jisu Huh, Ph.D., is Professor and Raymond O. Mithun Chair in Advertising at the Hubbard School of Journalism and Mass Communication, University of Minnesota – Twin Cities. Dr. Huh's research program has been centered on scientifically examining and better understanding advertising's positive and negative effects on consumers and its functional and dysfunctional roles in society. She directs the interdisciplinary Minnesota Computational Advertising Lab (https://mcal.umn.edu/), and specific areas of current research include computational advertising research, consumer trust and its role in advertising and information diffusion, advertising and consumer-brand engagement on social media, and direct-to-consumer advertising of healthcare products. Dr. Huh is Editor-in-Chief of the Journal of Advertising and the Past President of the American Academy of Advertising.
Dr. Steven Wu
Talk Title:
A Minimaximalist Approach to Reinforcement Learning from Human Feedback
Talk Description:
In this talk, we will introduce the Self-Play Preference Optimization (SPO) algorithm for reinforcement learning from human feedback (RLHF). The SPO approach is minimalist in that it does not require training a reward model or unstable adversarial training and is therefore rather simple to implement. SPO is also maximalist in that it provably handles non-Markovian, intransitive, and stochastic preferences while being robust to the compounding errors that plague offline approaches to sequential prediction. We utilize the Minimax Winner (MW) concept from social choice theory, which treats learning as a zero-sum game between policies. This allows a single agent to self-compete, enhancing convergence without dueling policies.
This is based on joint work with Gokul Swamy, Christoph Dann, Rahul Kidambi, and, Alekh Agarwal that will appear at ICML 2024. The paper is available at: https://arxiv.org/abs/2401.04056
Speaker Bio:
Dr. Steven Wu is an Assistant Professor in the School of Computer Science at Carnegie Mellon University, with a primary appointment in the Software and Societal Systems Department (with the Societal Computing program), and affiliated appointments with the Machine Learning Department and the Human-Computer Interaction Institute. He is also affiliated with the CyLab and the Theory Group.
His broad research interests are in algorithms and machine learning. These days he is excited about: Foundations of responsible AI, with emphasis on privacy, bias, and uncertainty considerations. Interactive learning, including contextual bandits and reinforcement learning, and its interactions with causal inference and econometrics. Economic aspects of machine learning, with a focus on learning in multi-agent settings.
Dr. Wu's research has been generously supported by the National Science Foundation (NSF) (including an NSF CAREER Award), the Okawa Foundation, CMU’s Block Center for Technology and Society, an Amazon Research Award, a Google Faculty Research Award, J.P. Morgan Faculty Awards, Meta Research awards, a Mozilla Research Grant, Apple, and Cisco Research.
Dr. Jie Ding
Talk Title:
Advancing Scalable AI: From Core Principles to Modern Applications
Talk Description:
In the rapidly evolving realms of data and learning ecosystems, the utility of artificial intelligence hinges on its ability to scale and adapt. This talk will delve into our recent research aimed at overcoming the challenges of catastrophic forgetting in continual learning. Distinct from traditional learning approaches, our approach emphasizes growing capability that enables learners to maintain performance on old tasks while integrating new information, rapidly adapt to new environments through the recollection of past knowledge, and actively solicit side information to accelerate learning. I will present novel theoretical underpinnings and practical algorithms with applications to large model training.
Speaker Bio:
Jie Ding is an Associate Professor at the School of Statistics, University of Minnesota. He received a Ph.D. in Engineering Sciences from Harvard University and a Bachlor's degree from Tsinghua University. Jie's research is at the intersection of AI, statistics, and scientific computing, with current focuses on the scalability and safety of large models. He is a recipient of the NSF CAREER Award, ARO Young Investigator Award, Cisco Research Award, and Meta Research Award.
Dr. Jonathan Bentz
Talk Title:
Generative AI, The Next (Re|E)volution in Computing
Talk Description:
Generative AI has taken the computing world by storm. In this talk we'll explore how computing has evolved and how the Generative AI explosion has pushed computing to new limits, and we'll discuss how NVIDIA's continued innovation in accelerated computing is helping to propel the field forward in ground-breaking ways.
Speaker Bio:
Jonathan Bentz is a Senior Solutions Architect at NVIDIA, leading a team of scientists and engineers focused on research computing customers. Jonathan and his group work as technical resources for universities, to support and enable their use of accelerated computing in HPC and AI. Prior to NVIDIA, Jonathan was a software engineer at Cray in the scientific libraries group. Jonathan obtained a PhD in physical chemistry and a MS in computer science from Iowa State University.
Dr. Ben Lynch
Talk Title:
Supporting AI Applications at the Minnesota Supercomputing Institute
Talk Description:
The Minnesota Supercomputing Institute (MSI) supports scientific computing at the University of Minnesota. This talk will give an overview of the state-of-the-art hardware and expert staff resources dedicated to facilitating cutting-edge AI projects. Attendees will learn how these resources can be leveraged to enhance their research outcomes, streamline workflows, and drive innovation in using AI technologies.
Speaker Bio:
Benjamin Lynch serves as the Director of the Minnesota Supercomputing Institute (MSI) at the University of Minnesota. In this capacity, he oversees the institute's operations, development, and research informatics activities, working closely with various leaders within the University system and other computing centers to support the research needs of the faculty. Ben's academic background is in computational chemistry, where he applied a variety of machine learning techniques in the development of wave-based and density-based electronic structure methods used in thermochemistry and thermochemical kinetics. He's an active member of the Coalition for Academic Scientific Computing,(CASC) and his current research interests are in the application of AI tools across a broad set of disciplines.
David Maeda
Talk Title:
Generative A.I. and Election Administration
Speaker Bio:
David Maeda was appointed to be the Director of Elections for the Office of the Secretary of State In January 2019. Prior to his appointment, David served 12 years as the Minnetonka 12 years as the Minnetonka City Clerk. He has also served as an elections supervisor for Washington and Hennepin Counties. David is a member of the National Association of State Election Directors and currently is the vice chair of the Electronic Registration and Information Center (ERIC). He also serves on the Humphrey School of Public Affairs Election Administration Advisory Board.
Eran Kahana
Talk Title:
Data Quality Control in AI: Navigating the AI Data Stewardship Framework.
Talk Description:
Data is the lifeblood of generative AI applications. And while these applications depend on access to enormous amounts of it, that is not enough. At this point we can see the relationship between quality and quantity: Data quality and quantity are equally important; scarcity in one inevitably destabilizes the other. The dynamics of this relationship becomes most evident by the performance of these applications, they are ultimately only as good as the data they train on. The AI Data Stewardship Framework (AI-DSF) is a framework designed to ensure high quality data is continuously provided. This talk will walk-through the AI-DSF, its purpose, selected use cases, and explain how the framework’s various controls collaborate to ensure the provision of high quality data.
Speaker Bio:
Eran Kahana is an experienced attorney concentrating his practice on cybersecurity, artificial intelligence, and intellectual property law. Eran is a fellow at Stanford Law School, a member of the advisory board of the Stanford Artificial Intelligence Law Society, an adjunct professor of law at the University of Minnesota Law School, a member of the Scientific Council of the Israeli Association for Ethics in Artificial Intelligence, and a co-author of The Law of Artificial Intelligence and Smart Machines, a publication of the American Bar Association.
Dr. Claire Segijn
Talk Title:
Ethical side-effects of dataveillance and GenAI
Talk Description:
Claire M. Segijn (Ph.D., University of Amsterdam) is an Associate Professor at the Hubbard School of Journalism and Mass communication and a Mithun Program Fellow in Advertising at the University of Minnesota.
Speaker Bio:
Claire M. Segijn (Ph.D., University of Amsterdam) is an Associate Professor at the Hubbard School of Journalism and Mass communication and a Mithun Program Fellow in Advertising at the University of Minnesota.
Dr. David Little
Talk Title:
Generative AI and the Electronic Health Record
Talk Description:
Objectives:
To understand the conceptual basis for deploying Generative AI in health care.
To demonstrate the current state of Generative AI in the electronic health record.
To examine early outcomes of Generative AI usage in the electronic health record.
To explore future directions for enhanced application of Generative AI in health care.
Speaker Bio:
Dave Little is the Director of Clinical Informatics at Epic in Verona, WI. A Family Physician by training, Dave practiced and taught at Wright State University Boonshoft School of Medicine for 19 years prior to joining Epic in 2010. His primary role at Epic is to support health care organizations in implementing and using the software successfully. He serves as Physician Liaison for 47 Epic organizations, and has supported over 60 go-lives since joining Epic. He is actively engaged in the Epic Opioid Management Workgroup and the Suicide Prevention Army. He also consults regularly with Epic developers regarding the clinical aspects of the software with a focus on the ambulatory applications. Most recently, he has worked with the Epic Health Research Network as an investigator and reviewer and has co-authored over 30 Research Briefs on EpicResearch.org.
Dr. Louis Kazaglis
Talk Title:
Generative AI and the Electronic Health Record
Talk Description:
Objectives:
To understand the conceptual basis for deploying Generative AI in health care.
To demonstrate the current state of Generative AI in the electronic health record.
To examine early outcomes of Generative AI usage in the electronic health record.
To explore future directions for enhanced application of Generative AI in health care.
Speaker Bio:
Louis Kazaglis is a Clinical Informatics Physician at Epic in Verona, WI. Prior to this he was a sleep physician and clinical informaticist at M Health Fairview and then Cleveland Clinic. He has an interest in improving the experiences and outcomes for physicians and patients alike, and leveraging technology to create scalable solutions for better healthcare for all.
Dr. Andrea Noel
Talk Title:
Generative AI and the Electronic Health Record
Talk Description:
Objectives:
To understand the conceptual basis for deploying Generative AI in health care.
To demonstrate the current state of Generative AI in the electronic health record.
To examine early outcomes of Generative AI usage in the electronic health record.
To explore future directions for enhanced application of Generative AI in health care.
Speaker Bio:
Dr. Andrea Noel is a physician on the Clinical Informatics team at Epic working closely with Epic’s software experts to create technology to improve the healthcare experience. She also serves as a liaison to healthcare organizations using Epic’s clinical products and a champion for physician well-being, efficient clinical workflows, and patient safety. Additionally, Dr. Noel acts as a clinical advisor for the development of predictive models, generative AI functionality, and clinical and research tools derived from observational data.
Dr. Shauna Overgaard
Talk Title:
Implementing quality management systems to close the AI translation gap and facilitate safe, ethical, and effective health AI Solutions.
Talk Description:
The integration of Quality Management System (QMS) principles into the life cycle of development, deployment, and utilization of machine learning (ML) and artificial intelligence (AI) technologies within healthcare settings holds the potential to close the AI translation gap by establishing a robust framework that accelerates the safe, ethical, and effective delivery of AI/ML in day-to-day patient care. Healthcare organizations (HCOs) can implement these principles effectively by embracing an enterprise QMS analogous to those in regulated industries. By establishing a QMS explicitly tailored to health AI technologies, HCOs can comply with evolving regulations and minimize redundancy and rework while aligning their internal governance practices with their steadfast commitment to scientific rigor and medical excellence.
Speaker Bio:
Dr. Shauna Overgaard leads the development and application of standardized approaches to create, test, implement, and monitor AI-based digital solutions, including developing the Mayo Clinic AI Evaluation and Documentation Framework. Dr. Overgaard established the Mayo Clinic AI Translation Advisory Board for multidisciplinary AI translation and enterprise integration and was recently appointed Co-Director of the Mayo Clinic AI Validation and Stewardship Program. She serves as an advisor on the Mayo Clinic SaMD Review Board for regulatory determination of AI solutions and on a national level as Co-Chair of the Coalition for Health AI Transparency Working Group, Co-Chair of the American Medical Informatics Association (AMIA) AI Evaluation Showcase, and Co-Chair of the National Academy of Medicine AI Code of Conduct Health Systems and Payers Working Group. Dr. Overgaard is Co-Director of the course Introduction to Deployment, Adoption & Maintenance of Artificial Intelligence Models/Algorithms in the Artificial Intelligence in Healthcare (AIHC) Graduate Program at Mayo Clinic College of Medicine and Science. Before her work in AI translation, she was involved in clinical research focused on the development of diagnostic tests leveraging principles of graph theory and multimodal data in the realm of neuroimaging, proteomic, and genomic data. Through this work, she maintains a responsibility to help facilitate national consensus-building efforts related to medical informatics and health AI.
Dr. Serguei Pakhomov
Talk Title:
ANNA: Automated Neural Nursing Assistant for Intensive Monitoring of Neurotoxicity
Talk Description:
We present a fully automated conversational AI based system for intensive monitoring of cognitive symptoms of neurotoxicity that frequently appear as a result of immunotherapy of hematologic malignancies. Early manifestations of these symptoms are evident in the patient’s speech in the form of mild aphasia and confusion and can be detected and effectively treated prior to onset of more serious and potentially life-threatening impairment. We have developed the Automated Neural Nursing Assistant (ANNA) system designed to conduct a brief cognitive assessment several times per day over the telephone for 5-14 days following infusion of the immunotherapy medication. ANNA uses a conversational agent based on a large language model to elicit spontaneous speech in a semi-structured dialogue, followed by a series of brief language-based neurocognitive tests. In this talk, I will describe the motivation for creating ANNA, its design and functional characteristics and share some of the challenges encountered so far with ANNA's deployment in an on-going prospective pilot study.
Speaker Bio:
Dr. Pakhomov is a Professor at the University of Minnesota College of Pharmacy. He has been working in the field of health informatics for over 20 years. In addition to a doctorate degree in Linguistics with a Cognitive Science focus at the University of Minnesota in 2001, he has undergone training in Medical Informatics and Clinical and Translational Science as an NIH Fellow at the Mayo Clinic. Dr. Pakhomov's research interests include evaluating and developing novel computational approaches to natural speech and language processing in the medical domain, including text of medical records and language produced by patients during cognitive testing. He studies spontaneous speech and language characteristics indicative of effects of medications and neurodegenerative disorders on human cognition, and works on developing natural language processing and machine learning techniques to extract information from text of clinical reports and biomedical literature. Most recently, he has been focusing on investigating deep learning approaches for automatic speech recognition and large language models to support the use of conversational agents in healthcare applications.
Dr. Hari Trivedi
Talk Title:
Ethics of Large Language Models in Medical Research
Talk Description:
Large Language Models (LLMs) hold transformative potential in medical research, yet they present ethical challenges, particularly concerning inherited biases and patient privacy. Biases embedded in training data can lead to inequitable healthcare outcomes, highlighting the necessity for robust bias mitigation strategies. Patient privacy is at risk as LLMs may inadvertently reveal sensitive information, necessitating stringent data protection measures. To uphold ethical standards, it is crucial to implement transparent practices, ensuring consistency, reliability, and accountability in the deployment of these advanced AI systems in healthcare.
Speaker Bio:
Dr. Scott Friedman
Talk Title:
Using NLP to Assess Stigma and Equity in Healthcare Documentation
Talk Description:
Health disparities exist for marginalized patient populations in the U.S., and recent findings suggest that stigmatizing language in clinicians' notes may worsen these disparities. We'll outline recent findings on the impact of language on clinicians' attitudes and care strategy, and we'll see how novel NLP approaches can identify various stigmatizing themes and positive themes in clinical notes in a context-sensitive fashion. We review findings that apply these tools to thousands of charts and correlate these linguistic themes with patient demographics.
Speaker Bio:
Scott Friedman, PhD is an AI and NLP specialist who leads research contracts for the DoD, NASA, and the U.S. Intelligence Community. He has published over 80 papers on AI, NLP, computational social science, human cognitive bias, and algorithmic bias. He is a Principal Research Scientist at SIFT and an independent AI consultant for healthcare systems.