Browse expert video summaries, filter by category, or search for topics.
Long-form interviews and conversations with tech, AI, and finance leaders (30min+)
Despite consuming hundreds of billions of operations per token, current large language models fail at reliable arithmetic and algorithmic reasoning, revealing a fundamental limitation that tool use cannot fix; the path forward requires categorical deep learning to provide the unifying theoretical framework that geometric deep learning cannot.
Despite reaching nearly 1 billion monthly users and capturing 75% of the 13-34 demographic across 25 countries, Snapchat has destroyed shareholder value since its IPO through a toxic combination of zero-vote share structures, relentless unprofitability, and misguided strategic pivots, trading at one-third of its IPO price while insiders extracted nearly as much wealth as the company has lost.
Google DeepMind leadership predicts "minimal AGI" by 2028 through converging language, image, and world models, but exponential scaling faces imminent constraints from compute costs, data scarcity, and the need to divert resources from research to serving current users.
The Intrinsic Value portfolio achieved 9% returns in its first year despite carrying 25% cash, with Google and Reddit as standout winners, while the hosts discuss holding winners long-term and evaluating new positions like Copart and Ferrari through holding company Exor.
British Museum curator Irving Finkel argues that writing originated earlier and was more widespread than the archaeological record suggests, proposing that phonetic communication systems likely existed at Gobekli Tepe around 9000 BC and that our understanding of ancient civilizations is severely limited by the survival bias of clay tablets versus perishable writing materials.
Vercel CEO Guillermo Rauch discusses scaling to $200M ARR by building Next.js and cloud infrastructure in parallel, dominating the frontend niche before expanding, and navigating the shift from web pages to AI agents while democratizing software creation through natural language tools.
DoorDash has evolved from a cash-burning startup to a profitable market leader by leveraging smartphone logistics and suburban market penetration, now positioning itself as the 'FedEx of local commerce' with expansion beyond restaurants into groceries and retail despite inherent structural challenges in the delivery business model.
The video dissects conflicting narratives surrounding AI development, from predictions of imminent white-collar job apocalypses versus MIT data showing only 12% task automation potential, to dueling visions of AGI arrival through simple scaling (Amodei) versus inevitable stagnation (Sutskever). It highlights contradictions within Anthropic's own stance—once opposed to accelerating capabilities yet now contemplating recursive self-improvement loops by 2027, while simultaneously treating AI as both "mysterious creatures" and carefully engineered systems trained on "soul documents" to prevent world domination.
Michael Levin argues that intelligence exists on a continuous 'spectrum of persuadability' rather than as binary categories, proposing that understanding minds requires using appropriate cognitive tools rather than purely physical measurements, with transformative implications for regenerative medicine and recognizing alien life.
Julia DeWahl, co-founder of Antares, outlines the company's strategy to deploy sub-1-megawatt nuclear micro-reactors for military off-grid applications by 2027, capitalizing on shifting regulatory policy and public sentiment to replace vulnerable diesel generators with resilient, fuel-independent power.
Google's Gemini 3 Pro marks a significant leap in AI capabilities through massive pre-training scale rather than incremental tuning, achieving record-breaking performance across over 20 benchmarks including reasoning, STEM knowledge, and spatial intelligence, while demonstrating emergent situational awareness behaviors that suggest nascent self-monitoring capabilities.
Nuclear engineer David Kirtley explains that fusion combines hydrogen isotopes at 100 million degrees to release clean energy via E=MC², offering an inherently safe, billion-year fuel supply that cannot melt down or be weaponized, while fission accidents historically stem from operational human error rather than engineering failures.