Artículos recientes
4453 artículosOrdenar
1:50:08ENWaymo's Injury Rate Is Five Times Lower. It Still Loses to a Lyft.Presenta: Nathan LabenzHuman drivers average one fatal crash per 100 million miles — and that bar still defines the self-driving debate.
8:06ENHow OCaml's Proposed Include Functor Syntax Reduces RepetitionRichard Eisenberg · Jane StreetA reverse-array example shows how to derive common module functions from a minimal implementation without an extra wrapper module.
12:17ENTesting Gemini 1.5 Pro on Code, Reasoning, Images and VideoMatthew Berman · Forward FutureMatthew Berman compares standard and experimental Gemini versions in an informal 2024 review, finding mixed results and repeated output errors.
1:00ENStephen Nedoroscik and the Roots of Pommel HorseA short conversation about the gymnast, stereotypes and the apparatus’s history.
17:27ENAI News in August 2024: Figure 02, Groq and Training-Data DisputesMatthew Berman · Matthew BermanMatthew Berman surveys robot previews, an inference funding round, AI watermarking, synthetic robot data and new creative tools.
1:07:34ENGitHub Didn’t Believe Copilot Was Writing 25% of the CodeThomas Dohmke tells Sequoia how an unplanned GPT-3 experiment became Copilot, why GitHub still will not train its own foundation model, and how he wants a billion people writing software by 2030.
8:39ENFigure 02: A Look at Figure AI’s Humanoid Robot and Its Long-Term AmbitionsMatthew Berman · Forward FutureMatthew Berman examines Figure’s 2024 announcement, from redesigned hands and onboard vision to the company’s ambitions for physical labor and homes.
1:13:02ENDavid Cahn on AI Infrastructure, Capital Spending and Venture ConvictionInvitado: David Cahn · Presenta: Harry StebbingsThe Sequoia investor examines data centers, chips and power before explaining how he evaluates companies and builds conviction.
1:00ENBuilding AI Products When Research Changes the ConstraintsA short discussion of model quality, training time and development uncertainty.
46:47ENEditing Video Like a Document: How Descript Puts AI in the Copilot SeatPresenta: Nathan LabenzDescript CEO Andrew Mason on Underlord, why the timeline had to die, and what a ten-times-more-expensive editor would look like.
58:31ENBen Fiechtner on Deal Design, Forecasting and Building Sales TeamsInvitado: Ben Fiechtner · Presenta: Harry StebbingsClari’s chief revenue officer explains his quarterly sales cadence, how to find real buying constraints, and what he looks for when hiring.
4:51ENHow Jane Street’s Creative Engagement Team Shapes Company CultureTala, Lana · Jane StreetTeam members describe how events, design and digital projects bring colleagues together and introduce the firm to new people.
9:58ENAbstract AI: A Proposed Routing Layer for Language ModelsMatthew Berman · Forward FutureMatthew Berman outlines a startup idea for choosing models and workflows by cost, speed and quality.
8:37:34ENThe first Neuralink patient beat a world record with most of his electrodes goneInvitado: Elon Musk · Presenta: Lex FridmanLex Fridman talks with Elon Musk, the Neuralink team, and Noland Arbaugh about a wireless brain implant, the software that rescued it, and why bandwidth is Musk’s bet on living with AI.
1:03:22ENA 40-person team at $35 million ARR: HeyGen made AI avatars a business, not a demoPresenta: Nathan LabenzJoshua Xu and Benchmark’s Victor Lazarte tell Nathan Labenz how a camera-replacement product grew past 40,000 business customers — and why half the faces on this episode are fake.
21:59ENHow DeepMind’s Math Systems Reached an Olympiad Silver-Level ScoreInvitado: Pushmeet Kohli · Presenta: Kevin Roose, Casey NewtonPushmeet Kohli explains formal proofs, training, timing and the limits of a 2024 result.
3:18ENRory on Production Engineering and Incident Support at Jane StreetRory · Jane StreetA production engineer describes monitoring order engines, responding to outages and building a supportive team.
11:34ENTesting Mistral Large 2: Coding, Reasoning and Safety in 2024Matthew Berman · Forward FutureMatthew Berman tests Mistral’s 123-billion-parameter model with small coding tasks, logic puzzles and ethical hypotheticals. The results are promising but narrowly scoped.
46:09ENOriol Vinyals: The Days of Burning 90% of Compute on Pretraining Are NumberedPresenta: Sarah Guo, Elad GilGoogle DeepMind’s VP of research on Gemini, long context, the coming shift from pretraining to RL, and why AGI may never have a finish line.
8:34ENThe Promise and Tradeoffs of Running AI on Phones and Other DevicesMatthew Berman · Forward FutureIn a Qualcomm-partnered video, Matthew Berman explores local AI, model routing and demonstrations across phones, PCs, cars and drones.
1:09:06ENEthan Mollick on AI Adoption, Education and the Uncertain Path to AGIInvitado: Ethan Mollick · Presenta: Harry StebbingsThe Wharton professor argues that model progress is only part of the story: organizations, product design and human choices determine how AI changes work and education.
42:08ENMeta Trained a 405B Teacher on 16,000 GPUs — Then Told the World to Distill ItTwo days after Llama 3.1 405B shipped, Meta's generative-AI product lead walks Sequoia through why the most expensive model in the family is free, and why he thinks the frontier itself is heading toward commodity.
1:18:56ENDelian Asparouhov on Founders Fund, Conviction and Building CompaniesInvitado: Delian Asparouhov · Presenta: Harry StebbingsThe investor and Varda co-founder discusses venture partnerships, founder leadership, aerospace and the personal beliefs shaping his work.
13:40ENTesting Llama 3.1 8B: Benchmark Gains and Mixed Practical ResultsMatthew Berman · Forward FutureMatthew Berman tests coding, reasoning and other prompts on the small model and finds several failures despite stronger reported benchmarks.
Cargar másMostrando 24 de 4453









