Image
Key Takeaways from Datapalooza 2025
Datapalooza 2025 explored how data is reshaping every facet of society. The School of Data Science welcomed more than 350 attendees in person to its flagship event, including students, faculty, alumni, industry partners, and community members, to its flagship event. Another 100 joined virtually.
This year's theme, Truth and Accountability in the Age of AI, examined the intersection of data science with other fields: algorithms, culture, design, health, national security, and the physical world. The breakout sessions framed current research in these areas and featured nearly 30 expert speakers.
Stephenson Dean Phil Bourne opened the event by reflecting on how Datapalooza, founded in 2015, continues to embody the School’s commitment to discovery and interdisciplinary impact. "Those of you who have been around a while might remember the phrase 'data is new oil, data is the new soil," Bourne reminisced. He noted data's value today, emphasizing the responsibility that data scientists have to embody "data science for good."
He went on to say that data science is no longer solely about models or algorithms. "It is about solving consequential problems in an era defined by geopolitical complexity, environmental change, and accelerating AI capabilities."
AI Mapping and Prediction in the Physical World
In the breakout session "Data and the Physical World," researchers demonstrated how data methods and computational tools are reshaping work in materials science, computer vision, and civil engineering.
Material scientist and Associate Professor Prasanna Balachandran shared how machine learning and large language models are accelerating materials discovery, turning what once required decades of laboratory experimentation into a streamlined computational workflow. His case study on the history of Teflon and PFA chemicals underscored how scientific discovery, industrial application, and environmental consequence form an intertwined narrative, one that today’s data tools might help prevent from repeating.
Assistant Professor of Data Science Lei Li followed with new insights into computer vision and generative AI, demonstrating models capable of reconstructing expansive 3D environments, simulating object interactions, and enabling designers to create and manipulate complex shapes using simple sketches. His work illustrated how perception, semantics, and prediction converge in systems that understand and interact with the world.
Machine learning scholar Florian Wiesner presented virtually and echoed Li's sentiments with a talk on the potential of a Physics Foundation Model, a type of specialized LLM that, if trained, could work out simulations of physics problems with real-world accuracy.
Professor of Civil Engineering Devin K. Harris closed the session by bringing data science down to the structural foundations of everyday life: bridges, roads, and large-scale infrastructure systems. Harris emphasized the reality of “data-scarce” environments, where sensors are limited, and inspections often rely on subjective human judgment. By deploying computer vision on video footage, his team can quantify structural cracking and deterioration, enabling more accurate predictions of infrastructure performance and preventing catastrophic failures.
AI as a tool for Cultural Analysis
The breakout session "Data and Culture" highlighted research at the intersection of environmental justice, cultural analysis, and the digital humanities.
Professor Lisa Blackmore introduced her concept of “tender data,” which reimagines environmental data not as abstracted metrics but as relational, community-centered narratives. Drawing from her work along the Bogotá River and with partner communities in Mexico and Brazil, she demonstrated how citizen-led data collection, artistic practice, and generative AI can counter the techno-scientific objectification of rivers.
Assistant Professor Adrienne Ghaly shifted the conversation to biodiversity loss, an ecological crisis that resists easy representation. She described a large-scale digital humanities project analyzing prose fiction from 1800 to 1925 to understand how culture has historically depicted the non-human world. The AI-driven project intends to map how biodiversity has been understood and written about over time through its reflections in literature.
Finally, Associate Professor of Data Science Rafael Alvarado described his research on Mayan cultural texts, underscoring AI's applications in the study of culture today. His research utilized simple algorithms to study the Popol Vuh, the Mayan story of creation, and produced his own suggested divisions for the book's chapters or acts.
Image
Keynote: Digital Authoritarianism Weaponizes Privacy
In a focused and urgent keynote, legal scholar Danielle Citron examined how privacy is increasingly used as a tool to undermine democratic norms. Citron argued that digital authoritarianism in the United States is emerging through selective privacy, where powerful actors shield themselves while exposing vulnerable communities.
She began by defining key terms, noting how concepts such as digital authoritarianism, harassment, and doxing are being distorted in public discourse. Citron showed how some federal employees receive heightened anonymity, including masked identification and limits on public recording, while others, often women, immigrants, or people of color, face online mobs, targeted watchlists, and the publication of personal information. These examples reveal a pattern in which privacy is treated as a privilege reserved for certain groups.
Citron warned that misusing terms like harassment and doxing weakens the ability of law enforcement to address genuine online abuse, including cyberstalking and nonconsensual image distribution. When political leaders apply these labels incorrectly, she said, it confuses the public, undermines First Amendment protections, and disrupts long-standing efforts to support victims.
Looking ahead, Citron called for stronger, technology-neutral laws that can adapt to rapid innovation. She emphasized the need for clear definitions, meaningful remedies, and collaboration among data scientists, policymakers, and legal experts. Protecting privacy in this moment, she said, requires understanding context, power, and harm, and ensuring that privacy safeguards support those most at risk.
Looking Ahead: AI Across the University
Interim Provost Brie Gertler offered closing remarks highlighting UVA’s ongoing efforts to support teaching, research, and public service in the age of AI. She described new faculty training programs, AI teaching resources, and institution-wide collaborations designed to help instructors adapt to a rapidly changing technological landscape. AI’s impact on learning, she noted, is not inherently positive or negative; its effects depend on how thoughtfully institutions engage with it.
Her message echoed a theme woven throughout Datapalooza 2025: AI is permeating every discipline, and the responsibility to guide its development rests with all of us.
