reSee.it - Related Video Feed

In this conversation, Lex Fridman speaks with Noam Brown, a research scientist at Facebook AI Research, who co-created AI systems that achieved superhuman performance in poker and the board game Diplomacy. Brown discusses the evolution of AI in games, particularly focusing on Libratus, which mastered heads-up No Limit Texas Hold'em, and Pluribus, which excelled in six-player poker. He emphasizes the significance of approximating Nash equilibrium in poker, where the AI's strategy of not adapting to opponents but rather playing optimally led to its success against top human players. Brown explains that No Limit Texas Hold'em differs from chess due to its high variance and the psychological aspects involved, such as bluffing and betting strategies. He notes that while both games reward strategic thinking, poker's unpredictable nature makes it more complex. The conversation also touches on the beauty of poker and the allure of finding an objectively correct way to play. The discussion transitions to Diplomacy, a game that combines strategy with negotiation and social dynamics. Brown highlights the challenges of creating an AI for Diplomacy, particularly due to the need for natural language processing and understanding human behavior. He explains that the AI must navigate complex social interactions and trust dynamics, making it distinct from purely adversarial games like poker. Brown describes the AI's training process, which involved self-play and leveraging human data to better understand human communication styles. He emphasizes the importance of trust in Diplomacy, noting that successful players often build alliances while managing the risk of betrayal. The AI, named Cicero, was able to perform competitively against human players, demonstrating its capability to negotiate and strategize effectively. The conversation also delves into the ethical implications of AI in games, particularly regarding deception and trust. Brown expresses excitement about the potential of AI to enhance our understanding of human interactions and decision-making processes. He suggests that the insights gained from AI in games like Diplomacy could inform broader applications in real-world scenarios, including geopolitics. Fridman and Brown conclude by discussing the future of AI, the challenges of data efficiency, and the philosophical questions surrounding AI's role in society. Brown encourages aspiring machine learning practitioners to embrace diverse perspectives and backgrounds, emphasizing the value of interdisciplinary approaches in tackling complex problems.

An idea from Steven Pinker reshapes how we read online upheaval: cancel culture is predictable because common knowledge guides collective punishment. Pinker argues Malcolm Gladwell’s cancellation was mathematically inevitable, rooted in a social media shaming mob. Common knowledge means I know something, you know it, I know that you know it, and so on. When a dissenting voice finally speaks, the room often echoes in agreement, illustrating a powerful coordination force that underpins civilization itself. Pinker's core distinction is between common knowledge and private or expert knowledge. The book stresses the difference between everyone knowing something and everyone knowing that everyone knows it. It introduces the idea of a shibileth or shibth, insider knowledge outsiders don’t share. When common knowledge falters—through disinformation or AI hallucinations—the foundations of cooperation wobble. Conspiracy theories and woo rise where communities lack a shared epistemic ground, and academia sometimes fears ideas that challenge established norms. Throughout the dialogue, censorship is framed as a tool to prevent common knowledge. Dictatorships and the Catholic Church suppressed demonstrations and the teaching that might unite believers. The Galileo episode shows that censorship targets not just speech but the spread of widely known ideas: Sidereus Nuncius was allowed, Dialogo was forbidden because it could coordinate dissent. Common knowledge thus becomes a weapon, while its suppression preserves power. Two pillars emerge: the Agree to Disagree theorem by Robert Aumann and the signaling logic in charity. If two rational agents share priors and their posteriors are common knowledge, they must converge; disagreement is unlikely with full information. In markets, prices reflect information because of common knowledge. In charity, publicly given gifts can signal virtue, while anonymous giving signals deeper altruism. The ladder of righteousness—from visible generosity to double-blind giving—shows how layers of mutual knowledge shape social rewards. An overarching thread ties to artificial intelligence. Large language models draw on vast text, computing patterns rather than grounding propositions in real-world knowledge. Pinker warns that hallucinations come from training data lacking a reliable knowledge base, producing a polluted form of common knowledge. The discussion also covers free will, determinism, and moral responsibility: even when brains operate under physical laws, people act as if they have free will to sustain social order, a tension that mirrors the puzzles of coordination in the book.

Reason, Lionel Page suggests, is less a tool for solving problems than a mechanism for convincing others. It’s why a courtroom argument often travels on clever framing rather than hard facts, and why our most constant debates are social tests rather than engineering challenges. He uses the 2001: A Space Odyssey image of a sudden flash of reasoning to illustrate how humans become human when we learn to bend information toward persuasion. Self-deception, he argues, is not a bug but a feature designed by evolution. We lie to ourselves to avoid costs, to bluff without appearing dishonest, and to preserve reputations. People consistently inflate how capable they are, how moral they are, and how victimized they have been, sometimes to secure a better share of resources or social status. The result is both a rose-tinted view of the world and a habit of arguing from the vantage point of the lawyer, not the scientist. From there the conversation moves to cooperation and conflict. Repetition makes trust possible because the future shadow of reputation discourages outright cheating. Language becomes a game of signals, where parents, partners, and coworkers negotiate through ambiguous statements, indirect asks, and paltering—the art of saying something true while steering others toward a false impression. Relevance, reciprocity, and a shared sense of belonging shape who succeeds and who stays outside the group, much as in a football match or a workplace project. Mind reading, theory of mind, and the social brain emerge as central concepts. Humans navigate nested beliefs, anticipate others’ moves, and regulate emotions to stay credible. The discussion pivots to artificial intelligence, with large language models offered as imitators of human conversation—impressive, but still far from the depth of genuine social understanding. Computers can simulate dialogue, yet they struggle with recursive mind reading and the subtle choreography of human cooperation. Ultimately, the episode reframes democracy as a contest of coalitions rather than a chase for universal truth. Leaders win by pleasing a shifting electorate, and loyalty signals—whether in politics, dating, or team sports—become as consequential as principles. The tension between autonomy and belonging remains a constant undercurrent, driving how we negotiate rules, punish betrayal, and invest in relationships. In Page’s view, acknowledging these games can cultivate more empathy and a healthier stance toward our own biases.

Welcome to the Huberman Lab Podcast. I'm Andrew Huberman, a Professor at Stanford, and today we discuss the biology, psychology, and utility of play. Play is essential not only for children but also for adults, as it enhances our nervous system and improves performance in various activities. Engaging in play allows exploration of new identities and fosters creativity, making individuals better leaders, workers, and learners. Research indicates that proper play can enhance focus and is being studied as a treatment for ADHD. Children lacking sufficient play are more prone to developing ADHD, but everyone can benefit from play at any age. I will share protocols and scientific insights that can help anyone, regardless of age, gain from play. A recent study published in Scientific Reports examined how reading on different devices affects comprehension and physiological responses. The study found that reading on smartphones leads to poorer comprehension compared to reading on paper. It also revealed that reading on smartphones suppresses physiological sighs, which are essential for oxygen intake and stress reduction. The prefrontal cortex becomes hyperactive when reading on devices, indicating a struggle to focus. To mitigate this, it's advisable to engage in physiological sighs regularly while reading on devices and to prefer reading from paper or larger screens. The study connects visual focus with autonomic function, suggesting that narrowing our visual window, as with smartphones, may inhibit our breathing patterns. Therefore, it's beneficial to broaden our visual window and read from larger screens or printed materials. Now, let's delve into play. Play is not just for children; it serves as a vital mechanism for testing and expanding our potential roles in various interactions. Play is homeostatically regulated, meaning that if we are deprived of play, we will engage in it more intensely when given the opportunity. Jaak Panksepp's research highlights that play is biologically essential, releasing endogenous opioids that enhance our cognitive flexibility. Play allows individuals to explore different roles and social dynamics in a low-stakes environment. This exploration is crucial for learning how to interact with others and understand social hierarchies. The various forms of play—social, individual, competitive—help us test outcomes and develop our identities. As we grow older, we often engage less in play, which can limit our creativity and adaptability. Engaging in play can enhance neuroplasticity, the brain's ability to change and adapt. The prefrontal cortex, responsible for executive function, becomes more flexible during play, allowing for exploration of new possibilities. To cultivate a playful mindset, it’s important to engage in activities that challenge you without high stakes. This could involve trying new sports, games, or even creative pursuits like chess, which encourages role adoption and strategic thinking. The key is to keep the stakes low and focus on the experience rather than the outcome. Personal play identity, a concept introduced by Gökhan Güneş, encompasses how we play, our personality, socio-cultural influences, and our environment. Reflecting on our childhood play experiences can reveal insights into our adult behaviors and preferences. In conclusion, play is a fundamental aspect of our biology that can enhance neuroplasticity and improve our overall well-being. I recommend dedicating at least one hour per week to engaging in pure play to reap these benefits. Play is not just about fun; it’s a vital tool for personal growth and development throughout our lives. Thank you for joining me in this exploration of play's power.

The Latitude Society in San Francisco lured members through a secretive invitation process involving immersive experiences. Participants received a card leading them to a mysterious building, where they underwent a series of dark, claustrophobic challenges, culminating in a storytelling session about an ancient society. The society, founded by Jeff Hull, aimed to create unique experiences but struggled financially, leading to controversial membership fees. Critics noted its cult-like traits, including isolation, information control, and emotional manipulation. Despite its intent, many members questioned if they had unwittingly joined a cult, highlighting the thin line between immersive experiences and cult dynamics.