Browse all summarized videos from our curated channels
Professor Michael I. Jordan critiques AGI as harmful PR that demoralizes young engineers, arguing that true intelligence requires economic and social systems thinking—treating billions of humans as agents in value-creating ecosystems—rather than isolated prediction engines built without intellectual foundations.
Pfizer CEO Albert Bourla discusses leveraging COVID-19 vaccine success to build organizational resilience, executing $80+ billion in strategic acquisitions targeting antibody-drug conjugates and obesity treatments, and restructuring R&D to double clinical success rates, while predicting AI will fundamentally reshape medicine within five years.
Journalist Megan Greenwald examines how private equity's incentive structure allows firms to profit through financial engineering and debt loading even when portfolio companies fail, creating devastating consequences for workers, communities, and industries ranging from retail to healthcare.
This lecture establishes that training data—not architecture—is the most critical and secretive component of modern language models, examining the technical impossibility of crawling the 'entire internet,' the three-stage data pipeline from raw web to specialized post-training, and the tightening legal constraints of copyright law and terms of service that increasingly restrict what can be legally used for training.
This lecture explores how to evaluate language models, examining different definitions of "good" from benchmark scores to economic usage, diving deep into perplexity as a foundational metric and its limitations, and tracing the evolution of exam-based benchmarks like MMLU from novel challenges to saturated metrics.
This lecture explores practical implementations of scaling laws for language models, focusing on the MiniCPM paper's techniques for stabilizing training across scales, including Maximal Update Parameterization (MUP) for consistent learning rates and Warm-up Stable Decay (WSD) schedules for efficient training continuation.
This lecture covers practical training of text-to-image diffusion models, detailing the evolution from UNet to Diffusion Transformer architectures, the three-stage training pipeline (pre-training, post-training, and tuning), and critical optimizations including flow matching loss functions, logit-normal time step sampling, and resolution-aware noise scheduling.
Stripe Sessions 2026 marked the company's most ambitious product launch day in history, centered on building economic infrastructure for the AI era. The keynote revealed a parabolic spike in new business formation since January 2026 and introduced tools including the Machine Payment Protocol, Link wallet for agents, and Stripe Projects to enable autonomous agent-to-agent commerce.
Sam Altman discusses the recent 'parabolic' inflection in AI capabilities, particularly for coding with GPT 5.5 and Codex, while outlining OpenAI's evolution into a massive-scale 'intelligence utility' provider focused on automating general computer work through agents like OpenClaw.
Nat Friedman and Daniel Gross describe the current era as the 'slow part' of the singularity, predicting that AI will drive massive economic shifts, force continuous security hardening, and enable a new golden age of personal hardware tinkering where AI agents reverse engineer proprietary systems in hours.
Stripe co-founder John Collison and Head of Data Emily Sans analyze 2026 economic data revealing that AI is driving unprecedented business formation and global scaling among lean solopreneurs, while commerce rapidly evolves toward autonomous agent-to-agent transactions using stablecoin micropayments.
The traditional Thucydides Trap doesn't apply to US-China relations because economic interdependence makes war catastrophic and meaningless, marking a shift from military-based to economics-based geopolitics.