Share
Foresight Institute Radio
Peter Norvig | A Modern Approach to AI
“I look at these assistants that we have today which are very primitive like Siri, Alexa and Google. But it's really interesting, because they represent a phase change in operating systems.”
Peter Norvig is a Director of Research at Google Inc. Previously he was head of Google’s core search algorithms group, and of NASA Ames’s Computational Sciences Division, making him NASA’s senior computer scientist. He received the NASA Exceptional Achievement Award in 2001.
In this episode, Peter Norvig speaks about how we can have a more modern approach to AI. Norvig starts with the history of the textbook. In 1990, the textbooks were subpar. AI was changing in three ways – moving from logic to probability, from hand coded knowledge to machine learning, and from duplicating human systems toward normative systems that got the best answer no matter what. After leaving Berkeley to go to Sun, he helped write a new textbook about AI.
Music:
I Knew a Guy by Kevin MacLeod is licensed under a Creative Commons Attribution 4.0 license. https://creativecommons.org/licenses/by/4.0/
Session Summary: Peter Norvig, Google | Ai: A Modern Approach - Foresight Institute
The Foresight Institute is a research organization and non-profit that supports the beneficial development of high-impact technologies. Since our founding in 1987 on a vision of guiding powerful technologies, we have continued to evolve into a many-armed organization that focuses on several fields of science and technology that are too ambitious for legacy institutions to support.
Allison Duettmann is the president and CEO of Foresight Institute. She directs the Intelligent Cooperation, Molecular Machines, Biotech & Health Extension, Neurotech, and Space Programs, Fellowships, Prizes, and Tech Trees, and shares this work with the public. She founded Existentialhope.com, co-edited Superintelligence: Coordination & Strategy, co-authored Gaming the Future, and co-initiated The Longevity Prize.
Apply to Foresight’s virtual salons and in person workshops here!
We are entirely funded by your donations. If you enjoy what we do please consider donating through our donation page.
Visit our website for more content, or join us here:
More episodes
View all episodes
Hacking Aging Biology | João Pedro de Magalhães
11:24|This recording with João Pedro de Magalhães is from Vision Weekend UK 2026, held in London in June 2026. View the recording along with the slides on our YouTube channel.João Pedro de Magalhães is Chair of Molecular Biogerontology at the University of Birmingham, following training at Harvard Medical School. He has published over 100 papers in journals including Science, Nature, and Cell, and maintains some of the most widely used databases in aging research, including GenAge, AnAge, and DrugAge.In this talk, João argues that aging is not a fixed biological constant and makes the case for a systematic scientific program to understand and intervene in the aging process. He covers the genomic and molecular data on long-lived species, the interventions that appear most promising, and where the field is likely to produce real breakthroughs in the years ahead.If you want to join a future Vision Weekend, details and upcoming dates are at foresight.org.The Case for AGI Safety Products | Marius Hobbhahn
10:30|This recording with Marius Hobbhahn is from Vision Weekend UK 2026, held in London in June 2026. View the recording along with the slides on our YouTube channel.Marius Hobbhahn is CEO and founder of Apollo Research, a London-based AI safety company that works with OpenAI, Google DeepMind, Anthropic, Microsoft, and Amazon to evaluate dangerous capabilities in frontier AI systems. Before founding Apollo, he completed a PhD in machine learning and worked on AI forecasting at EpochAI. Apollo's product, Watcher, provides real-time oversight and control for AI coding agents.In this talk, Marius makes the case that a specific category of AI safety research and tooling can both reduce the risks of AGI and be commercially viable. He argues that safety and profitability are not inherently in tension, and that the field needs more founders willing to build safety as a product rather than treat it purely as a research cost.If you want to join a future Vision Weekend, details and upcoming dates are at foresight.org.Situated, Universal AI Assistants | Greg Wayne
10:05|This recording with Greg Wayne is from Vision Weekend UK 2026, held in London in June 2026. View the recording along with the slides on our YouTube channel.Greg Wayne is Director of Research at Google DeepMind, where he leads Project Astra, Google's research prototype for a universal AI assistant. His research combines neuroscience with AI systems design, focusing on memory, contextual grounding, and how agents can reason over time.In this talk, Greg examines what it would take to build AI assistants that are truly situated: embedded in a physical and temporal context, capable of maintaining persistent memory, and able to act in the world rather than simply respond to queries. He draws on his work at DeepMind to outline the design challenges and the scientific questions that remain open.If you want to join a future Vision Weekend, details and upcoming dates are at foresight.org.It's Time to Harvest the Secure AI Tech Tree | Andrew Trask
51:27|This recording with Andrew Trask is from the Foresight Institute Secure AI seminar series, hosted in April 2026. Andrew Trask is the founder of OpenMined and creator of PySyft, one of the most widely used frameworks for privacy-preserving machine learning. He has spent the last decade building technical infrastructure that allows AI systems to train on sensitive data without exposing it. In this talk, Andrew makes the case that the field has developed a powerful set of privacy and security tools, including federated learning, homomorphic encryption, and differential privacy, and that the moment has arrived to move from research to real-world deployment. To see the slides that go along with this presentation, see the recording on YouTube. If you want to join future seminars live, the schedule is at foresight.org and in our Luma calendar.Innovation Needs More than Intelligence | Taylor Tondelli
14:37|This recording with Taylor Tondelli is from Vision Weekend Puerto Rico 2026, held in February 2026. In this talk, Taylor argues that intelligence alone, whether human or AI, is not what drives innovation. Raw capability needs the right conditions: incentives, institutional scaffolding, and the diversity of thought that lets new ideas find their footing. The talk examines what those conditions look like and where they tend to break down.Vision Weekend, where this talk was recorded, is where we gather leading builders, thinkers, and funders in AI, bio, nano, neuro, and space to celebrate technological progress and explore how to advance it for the betterment of life. If you want to join a future Vision Weekend live, upcoming events are listed at foresight.org and in our Luma calendar.The End of Money? | Neha Narula
13:01|This recording with Neha Narula is from Vision Weekend Puerto Rico 2026, hosted by the Foresight Institute in February 2026. Neha Narula is the Director of the MIT Digital Currency Initiative and a research scientist at the MIT Media Lab, where she has spent a decade studying digital currencies, payment systems, and the infrastructure of money. In this talk, she examines what the rise of digital currencies, central bank digital currencies, and programmable money means for the future of how value moves through the world, and what it would actually mean for money itself to change fundamentally.Vision Weekend, where this talk was recorded, is where we gather leading builders, thinkers, and funders in AI, bio, nano, neuro, and space to celebrate technological progress and explore how to advance it for the betterment of life. If you want to join a future Vision Weekend live, upcoming events are listed at foresight.org and in our Luma calendar.Mapping the Jagged Edges of AI | Tom Griffiths
54:54|This recording with Tom Griffiths is from our Secure AI seminar series, hosted in May 2026. Tom Griffiths is a professor of psychology and computer science at Princeton University and co-author of Algorithms to Live By, which explores how computational thinking can illuminate human judgment and decision-making. In this talk, Tom applies the tools of cognitive science to map the jagged edges of AI capability: the areas where AI dramatically outperforms human cognition, and the areas where it fails in ways human cognizers typically do not. He argues that understanding the precise shape of this frontier is essential for anyone building systems that combine human and AI judgment.If you want to join future seminars live, the schedule is at foresight.org and in our Luma calendar.Work, Love and Life When Robots Rule the Earth | Robin Hanson
11:36|This recording with Robin Hanson is from Vision Weekend Puerto Rico 2026, hosted by the Foresight Institute in February 2026. Robin Hanson is an economist at George Mason University and the author of The Age of Em: Work, Love and Life When Robots Rule the Earth. In this talk, he presents the core scenario of the book: a future in which whole brain emulations are the dominant form of intelligence, and works through the economic logic of what work, relationships, and social structure would look like under those conditions. It is a rigorous thought experiment that applies economic reasoning to questions that usually remain in the domain of science fiction.Vision Weekend, where this talk was recorded, is where we gather leading builders, thinkers, and funders in AI, bio, nano, neuro, and space to celebrate technological progress and explore how to advance it for the betterment of life. If you want to join a future Vision Weekend live, upcoming events are listed at foresight.org and in our Luma calendar.Fix the Money, Fix the World | Brandon Green
10:46|This recording with Brandon Green is from Vision Weekend Puerto Rico 2026, hosted by the Foresight Institute in February 2026. In this talk, Brandon Green makes the case that monetary infrastructure is upstream of many of the world's most persistent problems: that flawed money enables misalignment and short-termism at a systems level. He argues that fixing the foundations of how money works would unlock progress across a wide range of domains, and lays out what that would require.Vision Weekend, where this talk was recorded, is where we gather leading builders, thinkers, and funders in AI, bio, nano, neuro, and space to celebrate technological progress and explore how to advance it for the betterment of life. If you want to join a future Vision Weekend live, upcoming events are listed at foresight.org and in our Luma calendar.