UC Berkeley (Audio)

America’s contemporary democratic predicament is rooted in its historically incomplete democratization. Born in a pre-democratic era, the constitution’s balancing of majority rule and minority rights created still-unresolved dilemmas. Placing the U.S. in comparative perspective, Daniel Ziblatt, professor of government at Harvard University, offers new perspectives on what should be “beyond the reach of majorities” – and what should not – making the case for a fuller democracy as antidote to the perils of our age. Ziblatt is also director of the Transformations of Democracy group at Berlin’s WZB Social Science Center. He is the author of four books, including "How Democracies Die," co-authored with Steve Levitsky, a New York Times best-seller. His newest book co-authored with Steven Levitsky is entitled "Tyranny of the Minority." Series: "UC Berkeley Graduate Lectures" [Public Affairs] [Show ID: 39237]

What does it mean when we use the first-person pronoun ‘I’? And how does it relate to self-consciousness? In this program, Béatrice Longuenesse, professor of philosophy emerita at New York University, compares the analysis of philosophers Elizabeth Anscombe and Jean-Paul Sartre on consciousness, self-consciousness and the use of 'I'. Languenesse's current work spans the history of philosophy, especially Kant and nineteenth century German philosophy; the philosophy of language and mind; and philosophical issues related to Freudian psychanalysis. Series: "UC Berkeley Graduate Lectures" [Humanities] [Show ID: 39240]

Where did the American Dream of hard work equals upward mobility go? And what will it take to bring it back? In this talk, Raj Chetty, director of Opportunity Insights and professor of public economics at Harvard University, focuses on three policy levers to increase upward mobility: reducing racial and economic segregation through more effective affordable housing programs, investing in place-based policies, and strengthening higher education. Chetty gives specific examples of pilot studies and interventions that help inform the design of policy and practice from the federal to state to local levels, including at institutions of higher education such as UC Berkeley. He offers illustrations that can be scaled nationally, providing a pathway to expand opportunities for all. Series: "UC Berkeley Graduate Lectures" [Public Affairs] [Business] [Show ID: 39239]

Children’s chances of earning more than their parents have fallen from 90% to 50% over the past half century in America. How can we restore the American Dream of upward mobility for all children? In this talk, Raj Chetty, director of Opportunity Insights and professor of public economics at Harvard University, shows how big data from varied sources ranging from anonymized tax records to Facebook social network data is helping us uncover the science of economic opportunity. Among other topics, Chetty discusses how and why children’s chances of climbing the income ladder vary across neighborhoods, the drivers of racial disparities in economic mobility, and the role of social capital as a driver of upward mobility. He presents data on the state of economic opportunity in California in particular to provide a local context to these national patterns. Series: "UC Berkeley Graduate Lectures" [Public Affairs] [Business] [Show ID: 39238]

Young children’s learning may be an important model for artificial intelligence (AI). In this program, Alison Gopnik, professor of psychology and member of the Berkeley Artificial Intelligence Research (BAIR) Lab at UC Berkeley, says that comparing children and artificial agents in the same tasks and environments can help us understand the abilities of existing systems and create new ones. In particular, many current large data-supervised systems, such as large language models (LLMs), provide new ways to access information collected by past agents. However, they lack the kinds of exploration and innovation that are characteristic of children. New techniques may help to instantiate childlike curiosity, exploration and play in AI systems. This program is co-hosted with the UC Berkeley College of Computing, Data Science, and Society and the UC Berkeley Artificial Intelligence Research (BAIR) Lab. About the Series: CITRIS Research Exchange delivers fresh perspectives on information technology and society from distinguished academic, industry and civic leaders. Free and open to the public, these seminars feature leading voices on societal-scale research issues. Series: "Data Science Channel" [Science] [Show ID: 39351]

After founding four companies and working at top firms in venture capital and private equity, where fast growth and maximum profits rule, David Whorton, Founder and CEO of the Tugboat Institute, has spent the last decade exploring and developing the concept of the evergreen company—one built to last privately over 100 years. The evergreen company stands in contrast to those that are being built to flip to generate wealth for a small few. Instead, evergreen companies are being built with very long planning horizons and the commitment to share their success with their employees and their communities. Whorton argues evergreen companies are incredibly important to our society, but overlooked and under-appreciated relative to venture capitalists, private equity and public companies that represent the de facto growth company models. Since the dot.com boom, the de facto growth model for venture capitalists has been get-big-fast. It later evolved to growth-at-all-costs with the advent of cheap money under loose Fed policies. This play book led to numerous excesses, including the manic pursuit of ever larger and higher valuation rounds in hot companies. In the same period, private equity has risen dramatically, unwisely seen by many as a safer asset class than public stocks; an industry sits on over a trillion dollars of dry powder to invest, matched with a couple trillion of debt, giving the private equity firms purchasing power over $3 trillion dollars. Series: "Tanner Lectures on Human Values" [Business] [Show ID: 39235]

Researchers used to define objectives for artificial intelligence (AI) agents by hand, but with progress in optimization and reinforcement learning, it became obvious that it's too difficult to think of everything ahead of time and write it down. Instead, these days the objective is viewed as a hidden part of the state on which researchers can receive feedback or observations from humans — how they act and react, how they compare options, what they say. In this talk, Anca Dragan, Associate Professor of Electrical Engineering and Computer Sciences at UC Berkeley, discusses what this transition has achieved, what open challenges researchers still face and ideas for mitigating them. Dragan discusses applications in robotics and how the lessons there apply to virtual agents like large language models. Series: "Data Science Channel" [Science] [Show ID: 39350]

California’s deepest problems — the skyrocketing cost of housing, the lagging development of clean energy, the traffic choking the state — reflect an inability of Democratic governments to build real things in the real world quickly and affordably. The result is liberal governance that routinely fails to achieve liberal outcomes. New York Times opinion columnist and podcast host Ezra Klein talks with Amy E. Lerman, Chair and Professor of Public Policy and Political Science at UC Berkeley, about how we got here and what can be done about it. Series: "UC Berkeley Graduate Lectures" [Public Affairs] [Business] [Show ID: 39236]

In this program, Jaron Lanier, Microsoft's prime unifying scientist, discusses a piece he published in The New Yorker (“There Is No AI”) about applying data dignity ideas to artificial intelligence. Lanier argues that large-model AI can be reconceived as a social collaboration by the people who provide data to the model in the form of text, images and other modalities. This is a figure/ground inversion of the usual conception of AI as being a participant or collaborator in its own right. Explanations of model results and behaviors would then center around the relative influence of specific inputs through a provenance calculation mechanism. This formulation suggests new and different strategies for long-term economics in the context of high-performance AI, as well as more concrete approaches to many safety, fairness and alignment questions. This program is co-hosted with the UC Berkeley College of Computing, Data Science, and Society and the UC Berkeley Artificial Intelligence Research (BAIR) Lab. The CITRIS Research Exchange delivers fresh perspectives on information technology and society from distinguished academic, industry and civic leaders. Series: "Data Science Channel" [Science] [Show ID: 39326]

This episode of TecHype features Yoel Roth, former Head of Trust and Safety at Twitter. Yoel provides first-hand insights into how one of the largest online platforms in the world built out its trust and safety operations to better ensure its service was helpful, harmless, and aligned with user expectations While at Twitter, Dr. Roth found himself the target of a coordinated harassment campaign on the platform, one instigated by the current CEO Elon Musk. His years of work building out the trust and safety operations had become personal. In this episode, Dr. Roth provides professional and personal perspectives on the real benefits and risks of platform trust and safety efforts, the current state-of-the-art of the field, and where it’s going. Series: "UC Public Policy Channel" [Public Affairs] [Science] [Show ID: 39285]

Joan Donovan, a leading disinformation researcher specializing in media manipulation, explains how social media platforms have become the new battleground for public persuasion. Co-author of “Meme Wars: The Untold Story of the Online Battles Upending Democracy in America,” Donovan uncovers the ways memes and social media enable fringe groups to lure in new recruits and spread their ideologies. In this episode, Donovan provides expert guidance on technical and policy strategies necessary to mitigate the weaponization of social media. Series: "UC Public Policy Channel" [Public Affairs] [Science] [Show ID: 39286]

TecHype is a groundbreaking series that cuts through the hype around emerging technologies. Each episode debunks misunderstandings around emerging tech, provides insight into benefits and risks, and identifies technical and policy strategies to harness the benefits while mitigating the risks. This episode of TecHype features Prof. Stuart Russell from UC Berkeley, a world-renowned expert in artificial intelligence and co-author (with Peter Norvig) of the standard text in the field. We debunk misunderstandings around what “AI” actually is and break down the benefits and risks of this transformative technology. Prof. Russell provides an expert perspective on the real impacts AI will have in our world, including its potential to provide greater efficiency and effectiveness in a variety of domains and the serious safety, security, and discrimination risks it poses. Series: "UC Public Policy Channel" [Public Affairs] [Science] [Show ID: 39284]

UC Berkeley engineers have created a simple and low-cost new arsenic treatment system to help low-income communities access safer water. In many areas throughout California, the groundwater is tainted with dangerous levels of arsenic, a highly carcinogenic element that can seep into the water table from deposits in the soil and bedrock. While cities and larger municipalities can afford to remove arsenic from their water, many people living in small and rural communities are forced to choose between drinking contaminated tap water or purchasing bottled water — and those with private wells may not even know that their water is unsafe. (Video: Roxanne Makasdjian, Alan Toth, Adam Lau) Series: "UC Berkeley News" [Health and Medicine] [Show ID: 39226]

Researchers deployed a fleet of 100 semi-autonomous vehicles to test whether a new AI-powered cruise control system can help smooth the flow of traffic and improve fuel economy. In a massive traffic experiment, scientists tested whether introducing just a few AI-equipped vehicles to the road can help ease “phantom” jams caused by human behavior and reduce fuel consumption for everyone. (Video: Roxanne Makasdjian, Alan Toth, and CIRCLES Consortium Music: Dyalla - Organic Guitar House) Series: "UC Berkeley News" [Public Affairs] [Science] [Show ID: 39225]

TecHype is a groundbreaking series that cuts through the hype around emerging technologies to get to what matters. Each episode debunks misunderstandings around emerging tech, provides insight into benefits and risks, and identifies technical and policy strategies to harness the benefits while mitigating the risks of emerging technologies. This episode of TecHype features Prof. Hany Farid from UC Berkeley, a world-renowned expert in the analysis of digitally manipulated images. We take a deep dive into determining what a 'deepfake’ is and explore how these AI-generated images, videos, and audio can be used for both amusing and alarming purposes. Farid highlights the increasing prevalence of deepfakes and their impact on society. From revolutionizing the entertainment industry, bolstering creativity, and championing advocacy campaigns to their use in impersonating public figures in ways that manipulate elections or personal contacts to commit fraud. The episode concludes with a discussion of targeted strategies that can be pursued to keep you safe from harmful deepfakes, such as digital watermarking and detection tools. Series: "UC Public Policy Channel" [Public Affairs] [Science] [Show ID: 39244]

UC Berkeley drills a 400-foot borehole to explore geothermal heating on campus. UC Berkeley plans to decommission its 40-year-old cogeneration plant and replace its current steam heating system with a new system that uses water pipes to heat and cool buildings on campus. While the cogeneration plant burns natural gas to produce electricity and steam heat for the campus, the new system will use electricity for both power and thermal needs. By using clean energy sources, such as wind and solar, to produce this electricity, the campus’s future power, heating and cooling needs would be entirely carbon-free. (Video: Roxanne Makasdjian, Alan Toth, Adam Lau) Series: "UC Berkeley News" [Public Affairs] [Science] [Show ID: 39224]

Using real-life examples and historical evidence, French anthropologist Philippe Descola aims to understand the unique characteristics of communities that exist outside of modern societies. These communities have often been misunderstood because they were mistakenly compared to nation-states. However, Descola argues that we should examine the components and relationships within these communities based on how they perceive the world. By doing so, we can challenge the Eurocentric and human-centric view of social structures, which rely on Western ideas of progress and functionality. Series: "Tanner Lectures on Human Values" [Humanities] [Show ID: 38617]

In this program, scholars Philippe Descola, Adom Getachew, Timothy LeCain and David Wengrow discuss how views of humans verses non-humans shaped the modern world. Series: "Tanner Lectures on Human Values" [Humanities] [Show ID: 38618]

This series on artificial intelligence explores recent breakthroughs of AI, its broader societal implications and its future potential. In this presentation, Pamela Samuelson, professor of Law and Information at UC Berkeley, discusses whether computer-generated texts and images fall under the copyright law. She says that early on, the consensus was that AI was just a tool, like a camera, so humans could claim copyright in machine-generated outputs to which they made contributions. Now the consensus is that AI-generated texts and images are not copyrightable for the lack of a human author. The urgent questions today focus on whether ingesting in-copyright works as training data is copyright infringement and whether the outputs of AI programs are infringing derivative works of the ingested images. Four recent lawsuits, one involving GitHub’s Copilot and three involving Stable Diffusion, will address these issues. Samuelson has been a member of the UC Berkeley School of Law faculty since 1996. She has written and spoken extensively about the challenges that new information technologies pose for traditional legal regimes, especially for intellectual property law. She is a member of the American Academy of Arts & Sciences, a fellow of the Association for Computing Machinery (ACM), a contributing editor of Communications of the ACM, a past fellow of the John D. & Catherine T. MacArthur Foundation, a member of the American Law Institute, and an honorary professor of the University of Amsterdam. Series: "Data Science Channel" [Science] [Business] [Show ID: 38859]

This series on artificial intelligence explores recent breakthroughs of AI, its broader societal implications and its future potential. In this presentation, Sergey Levine, associate professor of Electrical Engineering and Computer Science at UC Berkeley, discusses AI reinforcement learning methods. Levine asks what it would take to create machine learning systems that can make decisions when faced with the full complexity and diversity of the real world, while still retaining the ability of reinforcement learning to come up with new solutions? He discusses how advances in offline reinforcement learning can enable machine learning systems to learn to make more optimal decisions from data, combining the best of data-driven machine learning with the capacity for emergent behavior and optimization provided by reinforcement learning. Levine received a BS and MS in Computer Science from Stanford University in 2009, and a Ph.D. in Computer Science from Stanford University in 2014. He joined the faculty of the Department of Electrical Engineering and Computer Sciences at UC Berkeley in fall 2016. His work focuses on machine learning for decision making and control, with an emphasis on deep learning and reinforcement learning algorithms. Applications of his work include autonomous robots and vehicles, as well as applications in other decision-making domains. His research includes developing algorithms for end-to-end training of deep neural network policies that combine perception and control, scalable algorithms for inverse reinforcement learning, deep reinforcement learning algorithms, and more. Series: "Data Science Channel" [Science] [Show ID: 38857]