On Tuesday, Karen Hao, New York Times Bestselling Author of Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI and recipient of the American Humanist Media Award in 2024, visited Charlottesville to share her insights and reporting experiences on the rapidly evolving AI industry since 2019. This event was presented by UVA’s Digital Technology for Democracy Lab, The Sloane Lab, and the College and Graduate School of Arts & Sciences.
The event began with Christa Acampora, Dean of the College and Graduate School of Arts and Sciences, welcoming the audience with an intriguing statement: “AI can save us.” While silence filled the space, she added, “Or destroy us.” Acampora began speaking about how AI “teaches transferable skills,” and can potentially open a path for us to “return to articulations that only we can do.” Her remarks set a timely tone, reminding audiences gathered in Old Cabell Hall, facing The School of Athens. “A particular image central for universal understanding,” that even in times of polarizing disagreements, thoughtful dialogues remain possible and valuable, promising an extraordinary discussion of the night.
Following the introduction, Karen Hao, Seth Lewis, and Mona Sloane took the stage. Seth Lewis, the incoming Elcan Jefferson Scholars Foundation Distinguished Professor of Artificial Intelligence and Media Studies, is a nationally recognized journalism and media scholar whose research focuses on the intersection of AI and journalism. Mona Sloane, an assistant professor of data science and media studies, aims to study design and inequality around AI development and regulations.
As the lighting on stage changes to a single spotlight surrounding the podium, Hao shared two excerpts from her book. The first, from Chapter 10, “Gods and Demons,” examines Silicon Valley’s and the Effective Altruism (“EA”) movement’s deterministic language when discussing the pursuit of artificial general intelligence (“AGI”). She argues that this framing treats innovation as though those driving it are not the authorities in control, while also revealing a lack of consensus on what AGI actually means.
This discussion, on how tech companies actively urge policymakers to focus on a deterministic, “inevitable” future of AI while downplaying the present harms caused by existing, non-frontier systems, has long been a central debate in the AI policy field. In 2023, Politico reported on researchers’ skepticism toward the motivations behind framing AI in terms of existential risk by “EA’s billionaire backers — who often possess close personal and financial ties to companies like OpenAI and Anthropic.” They argued that such narratives could serve as a distraction for lawmakers from immediate harms, including AI’s tendency to amplify racial and gender bias, erode privacy, and weaken copyright protections.
The second excerpt Hao shared was the prologue of her book. She described how, throughout her reporting on OpenAI, she recognized an urgent need for precise language to capture the magnitude of the economic and political power that companies like OpenAI have amassed, and why the word Empire best reflects this progression. Hao urged readers to question who creates AI models, who gets to shape them, and whose resources are being extracted in the process, while being cautious when being staged in front of the AI “illusion of progress.”
One recurring area in which questions about the authority behind AI deployment arise is the use of carceral and surveillance technologies. For example, in Gaza, Israeli forces have reportedly deployed secret facial recognition programs to monitor Palestinians and used AI models to generate lists of potential buildings to target in military strikes by analyzing data from surveillance, satellite imagery, and social networks. These AI-driven surveillance systems are then exported globally to police departments and militaries. As summarized in the report Prediction and Punishment, these models are “just one part of a complex of technologies fueled by longstanding geopolitical agendas of control, conquest, and exclusion.”
The discussion continued with Sloane posing a compelling question: “Can we imagine an AI outside of commerce?” To which Hao responded with the example of AI support in revitalizing te reo, the Polynesian language of the Māori people, the indigenous population of mainland New Zealand.
After British colonization of Aotearoa in 1840, English replaced te reo Māori, and the 1867 Native Schools Act banned its use in education. Urbanization further eroded Māori-speaking communities, dropping fluency from 90% to 12%. In 2016, as part of efforts to safeguard and increase accessibility of te reo, New Zealand’s Te Hiku Media, a broadcaster dedicated to te reo, employed AI tools to support its preservation and revitalization.
Using this example, Hao highlighted that AI models can be valuable without being generative. She emphasized the importance of community-centered, consent-based practices that ensure transparency and explainability in model development and use, with funding sourced from the community itself. Through this, Hao illustrated how AI can be deployed without commercial motives, used only when it genuinely happens to be the right solution. She further argued that “Silicon Valley’s one-size-fits-all approach simply doesn’t match the complexity of the world.”
As the event transitioned into the Q&A session, a journalist asked Hao about her research process and advice for future investigations in the AI field. Hao responded with an encouraging perspective, referencing the AI Spotlight Series she led design at the Pulitzer Center. She argued that technical expertise “cannot be the only expertise we cover,” urging inclusion of perspectives from politics, law, and other fields.
The event concluded with a long book-signing line in the Old Cabell Hall lounge as light rain drizzled outside. In a time of rapid transformation and expanding concentrations of information and power beyond public view, this evening with Karen Hao might have inspired a renewed sense of collective engagement, scaffolding a step toward more participatory decision-making in the future with AI.
Leave a Reply