Results
113 articlesSort
12:15EN
Matthew Berman· Jan 20, 2025DeepSeek-R1: Open-Weight Reasoning, Benchmark Results and a First TestMatthew Berman · Forward FutureMatthew Berman examines DeepSeek’s January 2025 release, compares reported benchmarks and prices, and tries a handful of reasoning prompts.
30:29ENMatthew Berman· Jan 3, 2025How Researchers Think o1-Style Reasoning Models WorkMatthew Berman · Forward FutureMatthew Berman explains a research survey of policy initialization, rewards, search and reinforcement learning, and the open questions behind reasoning models.
1:27:33EN
Cognitive Revolution "How AI Changes Everything"· Dec 3, 2024PodcastA 70B Model Booked Dinner 20% of the Time. MultiOn Took It to 95% in a DayHost Nathan LabenzMultiOn CEO Div Garg returns to The Cognitive Revolution to recap 18 months of agent hype, the AgentQ stack, and why 2025 might still be the year of vertical agents.
39:22EN
No Priors: AI, Machine Learning, Tech, & Startups· Nov 14, 2024Podcast · 2 speakersA Fields Medalist Stuck for Two Hours. AlphaProof Climbed for Three Days.Host Sarah Guo, Elad GilThe DeepMind team behind AlphaProof tells Sarah Guo and Elad Gil how an AlphaZero-style agent, a Lean verifier, and test-time reinforcement learning reached IMO silver — and why the missing skill is still theory building.
45:22EN
Sequoia Capital· Oct 2, 2024PodcastAlphaGo Thought 30 Seconds a Move. o1 Made “Think Longer” GeneralOpenAI researchers Noam Brown, Ilge Akkaya, and Hunter Lightman on teaching language models to reason, the generator-verifier gap, and a second scaling curve at inference time.
39:05ZH → English
Zhang Xiaojun Podcast· Sep 25, 2024Podcast · 2 speakersWang Xiaochuan on o1, Reinforcement Learning, and Slow ThinkingGuest 王小川 · Host 张小珺Baichuan Intelligence’s founder discusses possible technical paths behind o1, healthcare applications, and choices for AI startups.
39:45ENSequoia Capital· Sep 24, 2024PodcastHarmonic’s Bet: Teach AI Math Directly, and the Riemann Hypothesis Falls by 2029Robinhood’s Vlad Tenev and Harmonic co-founder Tudor Achim tell Sequoia why math is reasoning, why Lean is the verifier, and why they refuse to wait for mathematical skill to emerge from scale.
49:14ENSequoia Capital· Sep 17, 2024PodcastHumanoid robots fell from $1.5 million to $30,000. The GPT-3 moment is still a verbNvidia’s Jim Fan tells Sequoia how three kinds of data, a humanoid bet, and a System-1 motor-control model could unlock general-purpose robots.
51:15ENSequoia Capital· Sep 10, 2024PodcastAGI Has One Problem Left: Spend a Million Times More on a TokenMagic founder Eric Steinberger tells Sequoia’s Training Data why he trains his own models, why a software engineer is the MVP of AGI, and why 96% still is not a colleague.
1:27:02ZH → EnglishZhang Xiaojun Podcast· Sep 5, 2024Podcast · 3 speakersGuangmi: Self-Play RL Could Open a New Scaling Law for AGIGuest 广密 · Host 张小珺Guangmi believes the marginal gains from scaling language pretraining alone may be fading, making self-play reinforcement learning a more promising path. Whether it generalizes to coding, video, robotics, and beyond may determine the next stage of the AGI race.
50:33ENSequoia Capital· Aug 20, 2024PodcastThe 40% data-center save started with an email that ended in a question markPhaidra founder Jim Gao tells Sequoia how an AlphaGo-inspired bet at Google became reinforcement learning that runs billion-dollar plants.
46:09ENNo Priors: AI, Machine Learning, Tech, & Startups· Aug 1, 2024Podcast · 2 speakersOriol Vinyals: The Days of Burning 90% of Compute on Pretraining Are NumberedHost 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.
1:07:05ENSequoia Capital· Jul 16, 2024PodcastAlphaGo Combined Learning and Search. Language Models Only Did the First Half.Reflection AI's Misha Laskin tells Sequoia why prompted coding agents stall in the teens, and why the missing piece is still a ground-truth reward.
52:45ENCognitive Revolution "How AI Changes Everything"· Jun 13, 2024PodcastA $150 million model, $5 million six months later: Benchmark’s AI investment mathHost Nathan LabenzEric Vishria and Quilter’s Sergiy Nesterenko explain why PCB layout is not a Copilot problem, why human board data would be the wrong training set, and which AI companies look like sugar highs.
14:33ENMatthew Berman· May 1, 2024Astribot and Other Robot Demonstrations: Dexterity, Soccer and MobilityMatthew Berman · Forward FutureMatthew Berman reviews 2024 robot demonstrations, from Astribot's manipulation tasks to DeepMind soccer, Unitree flips and a legged monocycle.
56:56ENCognitive Revolution "How AI Changes Everything"· Apr 25, 2024PodcastA Week of PCB Layout in an Hour: Quilter Trains Agents, Not ImitatorsHost Nathan LabenzSpaceX alum Sergiy Nesterenko tells Will Summerlin how Quilter uses reinforcement learning to compile schematics into manufacturable boards, and why cloning human layouts would cap the system at human skill.
18:42ENMatthew Berman· Mar 20, 2024Robot Demonstrations in 2024: 1X, NVIDIA, Spot and BeyondMatthew Berman · Forward FutureMatthew Berman surveys humanoid, quadruped, drone and defense robotics, distinguishing company demonstrations from general-purpose autonomy.
9:15ENMatthew Berman· Mar 1, 2024Figure 01 and Other Robot Demos: What the 2024 Clips ShowMatthew Berman · Forward FutureMatthew Berman reviews a humanoid making coffee, inexpensive robot arms and other robotics demonstrations. The clips show specific tasks, not proof of artificial general intelligence.
1:43:47EN
Lex Fridman· Feb 16, 2024Podcast · 2 speakersYou Have to Run Before You Can Walk: Marc Raibert on Making Metal Look AliveGuest Marc Raibert · Host Lex FridmanThe Boston Dynamics founder tells Lex Fridman why aggressive hopping beat cautious walking, how 109 Atlas falls became a method, and why robots still need a second kind of intelligence.
1:58:43ENCognitive Revolution "How AI Changes Everything"· Feb 7, 2024PodcastA Worm With 302 Neurons Already Makes Trade-offs. AI Is Still on the Same Ladder.Host Nathan LabenzMax Bennett maps 600 million years of brain evolution onto five breakthroughs — and asks which rungs today's models have actually reached.
21:00ENMatthew Berman· Jan 23, 2024Jim Fan's Foundation Agent Vision: Skills, Robot Bodies, and SimulationMatthew Berman, Jim Fan · Forward FutureMatthew Berman examines research on Minecraft agents, varied robot bodies, and simulation as steps toward more general embodied AI.
35:16ENMatthew Berman· Nov 28, 2023What Was Known About Q* in 2023—and What Was SpeculationMatthew Berman, Sam Altman, Andrej Karpathy · Forward FutureMatthew Berman surveys rumors, reasoning papers and self-play analogies after OpenAI’s leadership crisis, with important limits on what was verified.
8:26ENMatthew Berman· May 28, 2023Voyager: How a GPT-4 Agent Learns Minecraft SkillsMatthew Berman · Forward FutureMatthew Berman explains a research system that proposes Minecraft tasks, stores reusable code and uses feedback to improve its next attempt.
6:53ENMatthew Berman· May 4, 2023StableVicuna’s 2023 Release: An Early RLHF Chatbot Put to the TestMatthew Berman · Forward FutureMatthew Berman reviews the 13-billion-parameter model and tests poetry, simple code, reasoning and a resignation email.Load moreShowing 24 of 113