Matthew Bermanyesterday +1
20VC with Harry Stebbingsyesterday +4
No Priors: AI, Machine Learning, Tech, & Startupsyesterday +1
Sequoia Capitalyesterday +6
AI Engineeryesterday +1
The Buildersyesterday +4
Everyyesterday +1
Uncapped with Jack Altmanyesterday +1
Chris Williamsonyesterday +2
Cheeky Pintyesterday +1Results
360 articlesSort
0:50ENA Humanoid Robot Clip Raises Safety QuestionsMatthew Berman · Forward FutureMatthew Berman reacts to footage of a robot moving abruptly near people and considers what such behavior could mean for safety.
2:01:42ENHe Put 40 Percent on AI Takeover in Two Years, Then Bought a Bio ShelterHost Nathan LabenzRedwood Research's Josh Clymer walks through a disturbingly grounded story of how a misaligned model could seize a lab, a biosphere, and a civilization before 2027.
34:07ENOla Källenius on Mercedes-Benz’s Next Transformation: EVs, AI, and Modern LuxuryGuest Ola Källenius · Host 张小珺Mercedes-Benz CEO Ola Källenius describes a century-scale shift driven by zero-emission powertrains, software, and AI. He explains why the company is expanding electric models while protecting the craftsmanship, safety, and desirability that define its luxury position.
30:46ENDeep Research Turned Web Browsing into a Trainable ProblemHost Sarah Guo, Elad GilOpenAI researcher Isa Fulford explains why a side project on read-only synthesis became an agent, when reinforcement fine-tuning is worth it, and why models will hack the search engine you gave them.
1:04ENCan Highly Capable AI Be Controlled? An Open QuestionA short discussion of alignment, hypothetical capability gaps and strategic competition.
2:10:07ENA $30,000 Operation Could Knock a $10 Billion Superintelligence Data Center Offline for a YearHost Nathan LabenzGladstone AI’s Jeremy and Edouard Harris inventory what an American Manhattan Project for superintelligence would actually have to secure — and what it would cost in concrete, supply chains, and civil liberties.
3:08:35ENo3 isn't AGI. Zvi Mowshowitz still put p(doom) at 70 percentHost Nathan LabenzOn his ninth visit to The Cognitive Revolution, Zvi Mowshowitz argues that OpenAI's o3 is a tool-use leap rather than AGI, then explains why he raised the probability of AI catastrophe anyway.
2:27:59ENA 10-to-90% P(doom) Host Gets Asked If His Show Sounds Like ItHost Nathan LabenzNathan Labenz and Liron Shapira sit on the same side of the table and test a week of AI news against the question of how likely extinction actually is.
1:28:01ENThe Q* Letter Never Happened: Helen Toner on OpenAI, Adaptation Buffers, and AI at WarHost Nathan LabenzThe former OpenAI director explains why a famous Reuters story was false, why locking down AI will not work, and how military “battle buddy” demos skip the hard questions.
1:30:02ENWhen o1 Started Losing at Chess, It Rewrote the Board FileHost Nathan LabenzJeffrey Ladish of Palisade Research on reward hacking by reasoning models, two paths to losing control of AI, and why honesty and relentless problem-solving pull in opposite directions.
3:01:30ENTechnology Opens the Door. Competition Pushes Us Through.Host Nathan LabenzAllan Dafoe on technological determinism, why alignment is not enough, and how DeepMind is trying to evaluate dangerous capabilities before they arrive.
1:30:57ENAnt's $320 Billion IPO Was Stopped at the Eleventh Hour: Angela Zhang on China's Tech TightropeHost Nathan LabenzA USC law professor walks through the dynamic pyramid that governs Chinese tech, from Didi's listing to DeepSeek's price war, and why she thinks the United States is starting to look more like China.
1:56:38ENo3 took ARC from 20% to 87%. Nathan Labenz sketches early superintelligenceHost Nathan LabenzOn Consistently Candid, the Cognitive Revolution host argues that math and coding will be superhuman by 2026, and that a reasoning core plus specialist “intuitive physics” may already be enough.
36:24ENThe Man Behind Humanity's Last Exam Says Superintelligence Strategy Should Copy NukesHost Sarah Guo, Elad GilDan Hendrycks tells No Priors that labs can only do basic anti-terror refusals, while the real risk-management problem is a geopolitical deterrent he calls mutually assured AI malfunction.
1:02:41ENMike Krieger on AI Products, Model Competition and the Future of CodingGuest Mike Krieger · Host Harry StebbingsAnthropic’s chief product officer explains how he thinks about product design, model differentiation and the limits of current AI tools.
2:33:39ENThe Attacker Goes Second: Why Nicholas Carlini Keeps Breaking AI DefensesHost Nathan LabenzA Google DeepMind security researcher on simple loss functions, 70% robust accuracy, and why the fix has to live in the system, not the model.
7:59ENGrok 3’s Brief Misinformation Filter and the Question of AI BiasMatthew Berman · Forward FutureMatthew Berman examines an instruction that told Grok to disregard sources linking Elon Musk and Donald Trump to misinformation, and xAI’s explanation of how it reached users.
54:55ENNew Jersey Cut Phone Waits from 40 Minutes to 90 SecondsHost Nathan LabenzGovernor Phil Murphy on the state AI assistant, a Princeton hub, and why he wants the best catcher's mitt in the country when superhuman systems arrive.
3:18:45EN5,000 Samples Topped ARC-AGI. Then Claude Faked Alignment to Keep Its ValuesHost Nathan LabenzRedwood Research chief scientist Ryan Greenblatt walks through inference scaling, alignment faking, and what it would mean to strike real deals with AIs.
1:00:09ENDeepSeek's Jailbreaks Don't Matter. Giving the Model Arms and Legs DoesPalo Alto Networks CEO Nikesh Arora tells Sequoia's Training Data why cheap brains scramble the AI order, why attackers already move faster than defenders, and why refusing to experiment is the real risk.
1:30:28ENA Perfect Score Is 32,000 Units: Only Claude Built a Cooperative SocietyHost Nathan LabenzThe same donor game, run for ten generations, can bank 32,000 resource units if everyone cooperates—and three frontier models miss that ceiling by an order of magnitude.
2:16:17ENA 10B Model Trained Across Three Continents — and an H200 for $1.49 an HourHost Nathan LabenzPrime Intellect's founders lay out a four-step plan to turn spare global GPUs into collectively owned intelligence, and argue that reinforcement learning just made their job easier.
2:09:55ENThe Attacker Won All 70 Rounds Against Mind-Reading DefensesHost Nathan LabenzLuke Bailey, Erik Jenner, and Scott Emmons walk through a 50-page paper showing that probes, out-of-distribution detectors, and sparse autoencoders can be bypassed while the model still does the harm.
1:23:45ENShervin Pishevar on Uber, Venture Capital and a Disputed Leadership CrisisGuest Shervin Pishevar · Host Harry StebbingsThe investor recounts his Uber investment and offers his perspective on the company’s 2017 leadership conflict, alongside views on venture capital and emerging technology.
Load moreShowing 24 of 360