Latest Interviews
Showing 781–795 of 2,162 transcripts.
Clear all filters- Sequoia Capital14 min
What's next for AI agentic workflows ft. Andrew Ng of AI Fund
AI agents are driving a paradigm shift from single-step prompting to iterative workflows that combine reflection, multi-agent collaboration, tool use, and planning to achieve results that can surpass larger, faster models running in zero-shot mode. This approach allows systems using smaller language models like GPT-3.5 to outperform GPT-4 on complex benchmarks such as HumanEval by enabling self-correction loops and specialized role delegation. While reflection patterns are now robust enough for immediate integration, emerging capabilities in planning and multi-agent debate are expected to dramatically expand the scope of autonomous tasks over the coming year.
- Dwarkesh Patel10 min
What a GPT-7 Intelligence Explosion Looks Like | Carl Shulman
The discussion outlines mitigation strategies for early detection of hostile AI motivations alongside a defined productivity threshold where AI contributions match or exceed human researcher output. It details operational mechanisms such as voting algorithms, cost-effective scaling of smaller models, and self-generated curricula that enable intelligence explosions without relying solely on brute force capability. These approaches combine hard physical constraints with empirical verification to ensure safety while accelerating innovation through distributed compute and structured learning environments.
- Dwarkesh Patel7 min
Sarah Paine – Maritime vs Continental Powers
The speaker contrasts Russia's antiquated continental model of territorial conquest with the maritime order's "win-win" system built on commerce and international law, arguing that Vladimir Putin's rejection of integration has squandered Russia's economic potential. By framing the current conflict as a strategic timeout for Russia, the analysis highlights how the Biden administration's multilateral approach has successfully mobilized former neutral nations like Finland and Sweden to enforce an impregnable border. Ultimately, the discourse advocates for preserving the post-WWII legal framework through a collaborative, non-hegemonic system that allows for Russia's eventual reintegration provided it adheres to established rules.
- Dwarkesh Patel8 min
Everyone Was Wrong About Intelligence – Dario Amodei (Anthropic CEO)
Industry experts acknowledge that commercial AI explosion timelines remain highly unpredictable, as current models display superhuman performance in constrained creative tasks while struggling with rigorous mathematical proof and multi-step reasoning. The event reveals that pre-training scaling has proven more efficient than reinforcement learning, yet a significant resource efficiency gap persists where synthetic intelligence processes vastly more data with far fewer synapses than the human brain without yet generating novel scientific breakthroughs. Despite this lack of new discovery, the speaker predicts that near-term improvements will enable these models to synthesize vast knowledge bases into new insights, particularly in biology where breadth of knowledge outweighs the derivation of new physical laws.
- Dwarkesh Patel7 min
Are We On Path Towards Superhuman Intelligence? – Dario Amodei (Anthropic CEO)
The speaker projects that economic investment and hardware advances will drive AI capabilities to match a generally educated human within two to three years, while acknowledging that scaling laws are currently bending to yield increasing returns. Despite this rapid acceleration, the speaker cautions against precise predictions of "superhuman" universality, noting that safety regulations and the complexity of physical embodiment may introduce significant messiness and delay. This trajectory suggests models will soon lead in scientific progress and specialized domains like math, yet the exact nature of their future impact remains distinct from traditional narratives of existential threat or total autonomy.
- Dwarkesh Patel7 min
How Did Dario & Ilya Know LLMs Could Lead to AGI?
OpenAI co-founders Ilya Sutskever and the speaker pioneered the hypothesis that general intelligence emerges by removing structural learning barriers and scaling data and compute rather than relying on specific domain engineering. Their research identified seven critical success factors, culminating in the Transformer architecture which eliminated context-window limitations and established language modeling as a substrate for complex reasoning. This strategy was validated by Alec Radford's GPT-1, confirming that massive scaling enables models to generalize across diverse tasks, a trajectory the founders view as having no inherent limits.
- a16z10 min
The Future of AI Is Amazing
The speaker argues that after seven decades of technical success, artificial intelligence has finally triggered a platform shift by eliminating the high marginal costs of creation and reasoning through foundation models. This economic transformation reduces the price of tasks like image generation and legal analysis by orders of magnitude, creating elastic demand that mirrors the historical impacts of the microchip and internet. Looking forward, this shift establishes a clear path toward economically viable embodied AGI, provided that venture capital, the tech sector, and Washington D.C. align their resources to support the necessary infrastructure and regulation.
- a16z19 min
Regulating Crypto in the Name of National Competitiveness
Coinbase COO Emily Arsenault characterizes the United States' lack of a clear crypto regulatory framework as a critical national security threat that has driven 70% of developers offshore and invited comparisons to the chip and 5G wars. In response to overzealous SEC and FTC actions, the company has partnered with former Defense Secretary Mark Esper to advocate for the FIT 21 bill while launching a campaign to highlight how blockchain technology inherently reduces illicit activity compared to cash. Arsenault argues that the current environment stifles innovation and M&A, urging policymakers to establish rules that secure the industry's future within US borders.
- a16z14 min
Securing America's Future: How Technology Companies and Washington Are Building A Safer World
Andoril, led by CEO Brian Schimpf, is advancing the U.S. Air Force's Loyal Wingman and Replicator initiatives by developing thousands of autonomous systems like the Roadrunner counter-drone to address modern air threats within a compressed two-year timeline. The company distinguishes itself through a free-market acquisition strategy and advanced commercial manufacturing, offering vertical takeoff and landing capabilities that counter expensive legacy interceptors while bypassing traditional defense acquisition bottlenecks. With active contracts from multiple U.S. government agencies and growing international demand, Andoril aims to establish strategic norms for autonomous warfare while solving critical operational gaps in air defense.
- a16z15 min
Collaborative Disruption at DoD: Kathleen Hicks in Conversation with Sharon Weinberger
Kathleen Hicks, Sharon Weinberger
Following a fatal drone attack on a U.S. vessel and observing Ukraine's rapid software iteration, the Department of Defense is accelerating the Replicator initiative to field thousands of autonomous capabilities by August 2025. To overcome bureaucratic bottlenecks and a continuing resolution, the DoD is expanding its industrial base to include commercial tech sectors and is seeking congressional approval for a record $100 billion fiscal 2024 budget to enable software-driven warfare. Key outcomes include the establishment of the Project Lima task force for generative AI and immediate system-level briefings for lawmakers to validate alternative acquisition pathways like Middle Tier Acquisition.
- a16z19 min
Rebuilding the California Dream
Jan, founder of California Forever, proposes constructing a walkable city for up to 400,000 residents on 60,000 acres in Solano County to address the Bay Area's housing crisis and stimulate defense manufacturing near Travis Air Force Base. Targeting a 2027 groundbreaking contingent on local voter approval, the project aims to offer homes at approximately $1 million and implement pedestrian-focused infrastructure to recapture lost GDP caused by housing constraints. By leveraging renewable energy and positioning the community within 1.5 hours of major tech hubs, the initiative seeks to reverse California's population decline and reunite Silicon Valley capital with the military-industrial complex.
- 80,000 Hours11 min
The next Black Death (or worse) | Most Pressing Problems
Phoebe Brooks, Luisa Rodriguez, Sarah Keeling, Jacob Sacks Jones, Alice Halstead, Alex Magill, Robert Stanton, Casey Grosso, Louisa Keight, Emma Pendarves, Lucrezia Pollice, Mikey Bligh-Smith, Ell Potter, June Archbold, Kamala Hussain, Oscar Galvin, Benedict Payne, Melissa Murphy, Lola Shaw, Felix Nugee, Bella Forristal, Cody Fenwick, Kempe Gay-Brydges, Anemone Franz, Benjamin Hilton, Nik Mastroddi, Arden Koehler
The presentation outlines the escalating global threat of pandemics arising from natural outbreaks, accidental laboratory leaks, and intentional bioweapon misuse, citing historical failures like the 1978 smallpox incident and the inadequacy of current containment systems. It argues that modern risks are compounded by advanced biotechnology and geopolitical instability, creating a potential for civilization-altering catastrophes that exceed historical precedents. To mitigate these dangers, the analysis advocates for a coordinated global coalition to invest in pathogen surveillance, adaptable vaccine platforms, and improved infrastructure rather than relying on blind hope.
- Jane Street11 min
Stack Allocation with Locals in OCaml | OCaml Unboxed
Jane Street is developing an open-source "local mode" branch for the OCaml compiler that reallocates data onto a specialized local stack to eliminate mark-and-sweep garbage collection pauses. By enforcing strict regional scoping where values cannot escape their lifecycle, this architecture enables constant-time deallocation and prevents the cache invalidation and stop-the-world latency inherent in standard heap allocation. The initiative aims to significantly boost low-latency performance by removing the overhead of garbage collection while maintaining memory safety through a novel mechanism called "exclaves."
- Y Combinator12 min
Stop Innovating (On The Wrong Things)
Dalton Caldwell, Michael Seibel
Founders are advised to channel their limited innovation energy into a single core value proposition to achieve product-market fit, avoiding the dilution of resources across multiple low-probability bets or contrarian structural choices. High-risk experiments in technical stacks, pricing models, and corporate governance often serve as anti-patterns that prioritize founder vanity over customer utility, a lesson illustrated by the failure of Digg compared to the focused success of Reddit. By deferring radical experimentation to a second venture, entrepreneurs can eliminate unnecessary friction and ensure their primary business hypothesis is the central driver of their early-stage strategy.
- Goldman Sachs9 min
Regional banks back in focus on real estate concerns
Ryan Nash of Goldman Sachs distinguishes current regional banking stress from the 2023 crisis by characterizing it as idiosyncratic and concentrated within smaller institutions rather than systemic, citing improved liquidity and accessible funding facilities. While commercial office loans and specific multifamily segments face multi-year headwinds, the industry holds adequate reserves and anticipates higher but manageable credit losses for 2024. Key risk exposure remains skewed toward banks under $150 billion in assets, though broader residential real estate dynamics and stabilizing net interest income support a positive outlook for sector growth.