Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Another Stanford interpretability guest lecture: Jack Merullo on "computational motifs" - the algorithmic primitives of transformers that show up again and again across circuits/tasks/models e.g. induction heads, binding vectors, helical representation comparisons, copy suppresion heads, etc. 00:53 - Intro: defining "computational motifs" 05:48 - Induction heads (a classic motif)... show more

Goodfire

24,260 subscribers

16,634 просмотров • 7 месяцев назад •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

NEW! vitalik.eth joins the greenpill.network podcast today to talk about Web3 Public Goods Funding in 2025. This episode is part 1 of a 2 part series. In ep 1 we discuss: TIMESTAMPS 00:00 - Intro 2:18 - Why Public Goods? 03:17- Private vs. Public Goods 05:20 - Challenges of Funding Public Goods 06:13 - Intrinsic Motivation and Public Goods 07:16 - Funding Models for Public Goods 09:18 - Digital Ecosystem and Public Goods 10:12 - Revenue Curve and Public Goods 11:36 - Decentralization vs. Domination 13:05 - Competitive Advantage of Public Goods 15:01 - Resilience Through Public Goods 17:07 - Broader Impact of Ethereum 19:18 - Escape Velocity Theory 19:51 - Importance of Public Goods 21:31 - Diversity in Funding Entities 23:44 - Challenges in Funding Public Goods 26:02 - Scaling Funding Needs 27:04 - Hybrid Funding Models 29:09 - Institutionalizing Funding 31:00 - Layer Two Solutions 31:59 - Importance of Scaling Funding 34:00 - Moralism in Ecosystem Dynamics 36:22 - Quality Allocation of Funding 37:26 - Diversity of Funding Mechanisms 39:34 - Stability in Funding Mechanisms 41:09 - Prediction Markets for Public Goods 42:27 - Info Finance Concept 44:36 - Distilled Human Judgment Mechanism 46:52 - Governance as a Lego Concept 47:58 - Forking Protocol Guild 50:54 - Discussion on Ethereum Critique 51:57 - Auto Public Goods Funding 53:11 - Tokenization and Open Source Funding 53:49 - Evaluating Funding Mechanisms 54:48 - Finding High Leverage Projects 56:19 - Challenges in Distribution 57:32 - Public Goods Funding in 2025 59:36 - Outro

NEW! vitalik.eth joins the greenpill.network podcast today to talk about Web3 Public Goods Funding in 2025. This episode is part 1 of a 2 part series. In ep 1 we discuss: TIMESTAMPS 00:00 - Intro 2:18 - Why Public Goods? 03:17- Private vs. Public Goods 05:20 - Challenges of Funding Public Goods 06:13 - Intrinsic Motivation and Public Goods 07:16 - Funding Models for Public Goods 09:18 - Digital Ecosystem and Public Goods 10:12 - Revenue Curve and Public Goods 11:36 - Decentralization vs. Domination 13:05 - Competitive Advantage of Public Goods 15:01 - Resilience Through Public Goods 17:07 - Broader Impact of Ethereum 19:18 - Escape Velocity Theory 19:51 - Importance of Public Goods 21:31 - Diversity in Funding Entities 23:44 - Challenges in Funding Public Goods 26:02 - Scaling Funding Needs 27:04 - Hybrid Funding Models 29:09 - Institutionalizing Funding 31:00 - Layer Two Solutions 31:59 - Importance of Scaling Funding 34:00 - Moralism in Ecosystem Dynamics 36:22 - Quality Allocation of Funding 37:26 - Diversity of Funding Mechanisms 39:34 - Stability in Funding Mechanisms 41:09 - Prediction Markets for Public Goods 42:27 - Info Finance Concept 44:36 - Distilled Human Judgment Mechanism 46:52 - Governance as a Lego Concept 47:58 - Forking Protocol Guild 50:54 - Discussion on Ethereum Critique 51:57 - Auto Public Goods Funding 53:11 - Tokenization and Open Source Funding 53:49 - Evaluating Funding Mechanisms 54:48 - Finding High Leverage Projects 56:19 - Challenges in Distribution 57:32 - Public Goods Funding in 2025 59:36 - Outro

owockai

213,967 просмотров • 1 год назад

At OpenAI, Chief Scientist Jakub Pachocki helps lead the research roadmap to AGI including a research intern-level AI system by September 2026 and a fully automated AI researcher by March 2028. I sat down with Jakub to check on those timelines and ask him all of my top-of-mind AI questions including: ▪️ How OpenAI thinks about extending RL beyond code and math ▪️ The current state of alignment research as more powerful models loom ▪️ The future of continual learning ▪️ How startups should think about building their own models/harnesses And he also shared some great stories around OpenAI’s pioneering work on math. YouTube: Spotify: Apple: 0:00 Intro 1:53 Research Intern Capability Timelines 4:59 Math Breakthroughs 7:59 RL Beyond Verifiable Tasks 12:32 RL vs In-Context 19:01 Allocating Compute Internally 28:18 AI for Science 31:40 Pattern Matching 33:23 Solving the Hardest Math Problems 37:40 Chain of Thought Monitoring 44:33 Generalization and Value Alignment in Models 47:57 Inside OpenAI 51:55 Quickfire

At OpenAI, Chief Scientist Jakub Pachocki helps lead the research roadmap to AGI including a research intern-level AI system by September 2026 and a fully automated AI researcher by March 2028. I sat down with Jakub to check on those timelines and ask him all of my top-of-mind AI questions including: ▪️ How OpenAI thinks about extending RL beyond code and math ▪️ The current state of alignment research as more powerful models loom ▪️ The future of continual learning ▪️ How startups should think about building their own models/harnesses And he also shared some great stories around OpenAI’s pioneering work on math. YouTube: Spotify: Apple: 0:00 Intro 1:53 Research Intern Capability Timelines 4:59 Math Breakthroughs 7:59 RL Beyond Verifiable Tasks 12:32 RL vs In-Context 19:01 Allocating Compute Internally 28:18 AI for Science 31:40 Pattern Matching 33:23 Solving the Hardest Math Problems 37:40 Chain of Thought Monitoring 44:33 Generalization and Value Alignment in Models 47:57 Inside OpenAI 51:55 Quickfire

Jacob Effron

138,223 просмотров • 3 месяцев назад

Why AI Progress Suddenly Feels Real - my conversation with Yann Dubois, who co-leads the Post-Training Frontiers team at OpenAI 00:00 - Intro 01:30 - Why recent AI progress feels like a step function 04:13 - Model reliability & the emotional rollercoaster of shipping GPT-5.5 07:33 - How OpenAI structures vertical and horizontal teams 09:49 - Improving model efficiency and test-time compute 12:32 - Yann's journey from Switzerland to OpenAI 15:37 - Reasoning in 2026: Real-world utility vs verifiable rewards 18:34 - GPT-5.5 Thinking vs Pro: Scaling test-time compute 20:09 - How reasoning models become more efficient 23:23 - Pre-training scaling and overcoming the data wall 27:03 - Multimodal data, synthetic data, and embodied AI 31:05 - Demystifying mid-training and post-training 37:21 - Does RL create new capabilities in AI? 38:53 - The challenges and frontier of scaling RL 43:09 - Is building AI models a craft or a strict science 48:21 - How AI models generalize across different domains 54:18 - How reinforcement learning cures AI hallucinations 56:04 - Negative generalization and conflicting instructions 58:05 - Can RL scale to law, medicine, and the broader economy? 1:00:19 - The evaluation bottleneck and Model as a Judge 1:04:21 - Continuous AI progress & continual learning 1:08:49 - Will foundation models eat the agent harness 1:11:23 - Why startups should focus on the last mile of AI

Why AI Progress Suddenly Feels Real - my conversation with Yann Dubois, who co-leads the Post-Training Frontiers team at OpenAI 00:00 - Intro 01:30 - Why recent AI progress feels like a step function 04:13 - Model reliability & the emotional rollercoaster of shipping GPT-5.5 07:33 - How OpenAI structures vertical and horizontal teams 09:49 - Improving model efficiency and test-time compute 12:32 - Yann's journey from Switzerland to OpenAI 15:37 - Reasoning in 2026: Real-world utility vs verifiable rewards 18:34 - GPT-5.5 Thinking vs Pro: Scaling test-time compute 20:09 - How reasoning models become more efficient 23:23 - Pre-training scaling and overcoming the data wall 27:03 - Multimodal data, synthetic data, and embodied AI 31:05 - Demystifying mid-training and post-training 37:21 - Does RL create new capabilities in AI? 38:53 - The challenges and frontier of scaling RL 43:09 - Is building AI models a craft or a strict science 48:21 - How AI models generalize across different domains 54:18 - How reinforcement learning cures AI hallucinations 56:04 - Negative generalization and conflicting instructions 58:05 - Can RL scale to law, medicine, and the broader economy? 1:00:19 - The evaluation bottleneck and Model as a Judge 1:04:21 - Continuous AI progress & continual learning 1:08:49 - Will foundation models eat the agent harness 1:11:23 - Why startups should focus on the last mile of AI

Matt Turck

100,966 просмотров • 2 месяцев назад

"Projects like the New Deal, the Apollo program pale in comparison to what we're doing right now." 🆕 Greg Brockman (Greg Brockman) joins us to talk GPT-5, GPT-OSS, and what's next on OpenAI's road to crystallizing all of human intelligence! “Energy turns into compute, turns into intelligence… crystallizing compute into potential energy you can release again and again.” 0:00:04 - Introductions 0:01:04 - The Evolution of Reasoning at OpenAI 0:04:01 - Online vs Offline Learning in Language Models 0:06:44 - Sample Efficiency and Human Curation in Reinforcement Learning 0:08:16 - Scaling Compute and Supercritical Learning 0:13:21 - Wall clock time limitations in RL and real-world interactions 0:16:34 - Experience with ARC Institute and DNA neural networks 0:19:33 - Defining the GPT-5 Era 0:22:46 - Evaluating Model Intelligence and Task Difficulty 0:25:06 - Practical Advice for Developers Using GPT-5 0:31:48 - Model Specs 0:37:21 - Challenges in RL Preferences (e.g., try/catch) 0:39:13 - Model Routing and Hybrid Architectures in GPT-5 0:43:58 - GPT-5 pricing and compute efficiency improvements 0:46:04 - Self-Improving Coding Agents and Tool Usage 0:49:11 - On-Device Models and Local vs Remote Agent Systems 0:51:34 - Engineering at OpenAI and Leveraging LLMs 0:54:16 - Structuring Codebases and Teams for AI Optimization 0:55:27 - The Value of Engineers in the Age of AGI 0:58:42 - Current state of AI research and lab diversity 1:01:11 - OpenAI’s Prioritization and Focus Areas 1:03:05 - Advice for Founders - It's Not Too Late 1:04:20 - Future outlook and closing thoughts 1:04:33 - Time Capsule to 2045 - Future of Compute and Abundance 1:07:07 - Time Capsule to 2005 - More Problems Will Emerge

"Projects like the New Deal, the Apollo program pale in comparison to what we're doing right now." 🆕 Greg Brockman (Greg Brockman) joins us to talk GPT-5, GPT-OSS, and what's next on OpenAI's road to crystallizing all of human intelligence! “Energy turns into compute, turns into intelligence… crystallizing compute into potential energy you can release again and again.” 0:00:04 - Introductions 0:01:04 - The Evolution of Reasoning at OpenAI 0:04:01 - Online vs Offline Learning in Language Models 0:06:44 - Sample Efficiency and Human Curation in Reinforcement Learning 0:08:16 - Scaling Compute and Supercritical Learning 0:13:21 - Wall clock time limitations in RL and real-world interactions 0:16:34 - Experience with ARC Institute and DNA neural networks 0:19:33 - Defining the GPT-5 Era 0:22:46 - Evaluating Model Intelligence and Task Difficulty 0:25:06 - Practical Advice for Developers Using GPT-5 0:31:48 - Model Specs 0:37:21 - Challenges in RL Preferences (e.g., try/catch) 0:39:13 - Model Routing and Hybrid Architectures in GPT-5 0:43:58 - GPT-5 pricing and compute efficiency improvements 0:46:04 - Self-Improving Coding Agents and Tool Usage 0:49:11 - On-Device Models and Local vs Remote Agent Systems 0:51:34 - Engineering at OpenAI and Leveraging LLMs 0:54:16 - Structuring Codebases and Teams for AI Optimization 0:55:27 - The Value of Engineers in the Age of AGI 0:58:42 - Current state of AI research and lab diversity 1:01:11 - OpenAI’s Prioritization and Focus Areas 1:03:05 - Advice for Founders - It's Not Too Late 1:04:20 - Future outlook and closing thoughts 1:04:33 - Time Capsule to 2045 - Future of Compute and Abundance 1:07:07 - Time Capsule to 2005 - More Problems Will Emerge

Latent.Space

305,090 просмотров • 11 месяцев назад

Today on MCG: | Marketing Collector Crypt Collector Crypt is the #1 consumer app on Solana right now, on pace for $85M in May revenue with $30M in inventory on-chain BassBuddah breaks down the ComicBook.com partnership (40M users), the Loopscale lending integration that lets you borrow against your Charizard, the Metaplex compressed NFT work, and where the token is headed once regulation catches up 👇 00:02 - Intro 01:40 - The King welcomed back 02:41 - Soft rebrand reveal, new logo 03:23 - ComicBook.com partnership 04:30 - The gotcha machine guarantees authenticity and quality 05:08 - The pitch 05:50 - Proving any collectible can be tokenized, not just TCG's 06:51 - "We are 5% loading"...Pokemon is 85% of revenue, One Piece 10%, sports 5% 07:18 - Splitting basketball, baseball, and football into their own gotcha machines 07:50 - Sports cards on a tear 09:35 - Loopscale partnership announced 11:05 - Borrow 40-75% of your Charizard's value in USDC, pay back, get the card back 13:15 - White-label thesis 15:36 - Slabz fanboy 17:07 - IRL vending machines as the next frontier 18:42 - Brick and mortar storefronts, grading partnerships, event activations on the roadmap 19:10 - Ebay and Gamestop tension 22:48 - $CARDS token primer 27:51 - $42K airdropped to $CARDS holders this month 30:33 - Compliance disclaimer 31:36 - "30 million in inventory" backed by the company, hopeful one day backed by the token 32:33 - DAT strategy 36:53 - Yield idea 37:39 - Marketplace update 39:58 - USDC offers, counteroffers, username profiles, in-app messaging 41:33 - One-of-a-kind escrow 42:06 - Lifetime buyback dashboard 43:02 - Trustless card-for-card trades, zero fee, any category for any category 45:53 - Why TCG projects support each other 48:43 - Why TCG's blew back up 51:39 - Buybacks confirmed as a priority 52:44 - "We are the number one consumer app on Solana right now" 53:50 - Why this is Solana RWA product-market fit 55:00 - LA Collecticon June 12 57:50 - Miami strategy 59:13 - Ansem connection clarified 1:01:24 - Future livestreams 1:06:29 - Closing analysis 1:08:42 - Takeaways 1:12:54 - Why isn't $CARDS on a CEX yet?

MCG

107,440 просмотров • 2 месяцев назад

Sugar Diet Strategies to Maximize Weight Loss In this episode, Joshua Rainer discusses the rising trend of sugar in diets, exploring its implications for health and performance. He shares his personal journey through various dietary approaches, emphasizing the importance of sugar for energy and recovery. Rainer critiques the cultural stigma surrounding sugar and advocates for a balanced approach to nutrition that includes adequate carbohydrates. He highlights the need for micronutrients and the role of metabolism in achieving health goals, ultimately promoting a more open-minded perspective on sugar consumption. 00:00 The Sugar Trend: A New Dietary Phenomenon 02:54 Dietary Shifts: From Low Carb to Sugar 05:49 Personal Journey: The Evolution of Joshua's Diet 09:12 The CrossFit Influence: Changing Perspectives on Carbs 12:00 The Role of Sugar in Performance and Recovery 14:53 Ray Peat and the Pro-Metabolic Approach 18:09 Cultural Acceptance of Sugar: A Shift in Mindset 21:02 The Intersection of Veganism and Sugar Consumption 23:54 The Metabolic Rate Dilemma: Low Energy Diets 26:45 The Honey Diet: A New Framework for Nutrition 30:10 Conclusion: Finding Balance in Dietary Choices 41:42 The Evolution of Human Diets 44:02 Performance and Nutrition: A Modern Perspective 45:59 The G-Flux Concept: Eating More to Burn More 48:05 Metabolism and Energy: The Role of Food 51:27 Cultural Shifts in Dietary Approaches 54:19 Micronutrient Awareness in High-Carb Diets 57:43 The Importance of Cholesterol and Vitamin A 01:01:39 Navigating Nutrient Deficiencies 01:04:05 The Role of B Vitamins in Metabolism 01:05:33 Cyclical Dieting: Balancing Nutrients 01:08:21 Fat Loss Strategies: Sugar and Lean Meat 01:11:45 Experimentation with Dietary Approaches 01:15:39 Final Thoughts on Nutrition and Performance

Sugar Diet Strategies to Maximize Weight Loss In this episode, Joshua Rainer discusses the rising trend of sugar in diets, exploring its implications for health and performance. He shares his personal journey through various dietary approaches, emphasizing the importance of sugar for energy and recovery. Rainer critiques the cultural stigma surrounding sugar and advocates for a balanced approach to nutrition that includes adequate carbohydrates. He highlights the need for micronutrients and the role of metabolism in achieving health goals, ultimately promoting a more open-minded perspective on sugar consumption. 00:00 The Sugar Trend: A New Dietary Phenomenon 02:54 Dietary Shifts: From Low Carb to Sugar 05:49 Personal Journey: The Evolution of Joshua's Diet 09:12 The CrossFit Influence: Changing Perspectives on Carbs 12:00 The Role of Sugar in Performance and Recovery 14:53 Ray Peat and the Pro-Metabolic Approach 18:09 Cultural Acceptance of Sugar: A Shift in Mindset 21:02 The Intersection of Veganism and Sugar Consumption 23:54 The Metabolic Rate Dilemma: Low Energy Diets 26:45 The Honey Diet: A New Framework for Nutrition 30:10 Conclusion: Finding Balance in Dietary Choices 41:42 The Evolution of Human Diets 44:02 Performance and Nutrition: A Modern Perspective 45:59 The G-Flux Concept: Eating More to Burn More 48:05 Metabolism and Energy: The Role of Food 51:27 Cultural Shifts in Dietary Approaches 54:19 Micronutrient Awareness in High-Carb Diets 57:43 The Importance of Cholesterol and Vitamin A 01:01:39 Navigating Nutrient Deficiencies 01:04:05 The Role of B Vitamins in Metabolism 01:05:33 Cyclical Dieting: Balancing Nutrients 01:08:21 Fat Loss Strategies: Sugar and Lean Meat 01:11:45 Experimentation with Dietary Approaches 01:15:39 Final Thoughts on Nutrition and Performance

Josh Rainer

111,552 просмотров • 1 год назад

E159: Hyperliquid: Housing all of Finance jeff.hl came back on the When Shift Happens Podcast to talk about the Hyperliquid journey since the TGE and what the future holds for one of the most loved and prolific protocols in the space Hyperliquid Timestamps 0:00 Intro 2:01 Singapore 2:27 Reminiscing on the Token Launch 5:00 Was This Scale Of Wealth Expected? 6:28 Doing The Right Thing In Crypto 9:07 The Responsibility that comes with Billions of $ 11:10 Jupiter KAST 11:51 Bringing Hyperliquid to the masses 15:21 Pre TGE and Post TGE: Operational difference 20:13 Choices on what to build Internally vs Externally 22:05 How to build a reliable team 24:51 Did the Team celebrate the HYPE wealth Generation event? 26:45 How to test talents for High Integrity 28:31 How much does the Hyperliquid team sleep? 30:05 Employee Vesting Fears 31:41 Dealing with FUD 32:28 How Does Jeff Personally Handle FUD 35:02 Token "Buybacks" critics 37:20 Why Hyperliquid can't have Discretionary "Buybacks" 39:04 HyperEVM, explained Simply 40:00 Paradex Zodl 40:41 HyperEVM: Success so Far? 44:05 HIP-3, explained Simply 47:44 What makes Hyperliquid's approach different 48:19 Why Should People Care? 51:33 Bring All Finance On Chain 52:08 Why Is The Hyperliquid Approach Better? 53:47 Key Numbers showing that Hyperliquid Is Doing it right 59:01 What Has the Unit team demonstrated with spot trading on Hyperliquid in 2025 1:03:29 HIP-4: Outcome Markets 1:08:01 Trezor Sui 1:08:58 What does "Housing All Of Finance" mean? 1:10:51 Why Hyperliquid is not a crypto company 1:12:23 Why Does Hyperliquid have A Stablecoin USDH (Native Markets) 1:14:39 What Is Kinetiq & Why Does It Matter? 1:16:15 Why Is What HyperLend Is Building Important For HyperLiquid 1:23:39 Where did Fairness cost the most? 1:24:47 What should Hyperliquid be Remembered for? 1:25:24 Why should people stay in Crypto when there's an AI brain drain? 1:28:10 Closing Thoughts

E159: Hyperliquid: Housing all of Finance jeff.hl came back on the When Shift Happens Podcast to talk about the Hyperliquid journey since the TGE and what the future holds for one of the most loved and prolific protocols in the space Hyperliquid Timestamps 0:00 Intro 2:01 Singapore 2:27 Reminiscing on the Token Launch 5:00 Was This Scale Of Wealth Expected? 6:28 Doing The Right Thing In Crypto 9:07 The Responsibility that comes with Billions of $ 11:10 Jupiter KAST 11:51 Bringing Hyperliquid to the masses 15:21 Pre TGE and Post TGE: Operational difference 20:13 Choices on what to build Internally vs Externally 22:05 How to build a reliable team 24:51 Did the Team celebrate the HYPE wealth Generation event? 26:45 How to test talents for High Integrity 28:31 How much does the Hyperliquid team sleep? 30:05 Employee Vesting Fears 31:41 Dealing with FUD 32:28 How Does Jeff Personally Handle FUD 35:02 Token "Buybacks" critics 37:20 Why Hyperliquid can't have Discretionary "Buybacks" 39:04 HyperEVM, explained Simply 40:00 Paradex Zodl 40:41 HyperEVM: Success so Far? 44:05 HIP-3, explained Simply 47:44 What makes Hyperliquid's approach different 48:19 Why Should People Care? 51:33 Bring All Finance On Chain 52:08 Why Is The Hyperliquid Approach Better? 53:47 Key Numbers showing that Hyperliquid Is Doing it right 59:01 What Has the Unit team demonstrated with spot trading on Hyperliquid in 2025 1:03:29 HIP-4: Outcome Markets 1:08:01 Trezor Sui 1:08:58 What does "Housing All Of Finance" mean? 1:10:51 Why Hyperliquid is not a crypto company 1:12:23 Why Does Hyperliquid have A Stablecoin USDH (Native Markets) 1:14:39 What Is Kinetiq & Why Does It Matter? 1:16:15 Why Is What HyperLend Is Building Important For HyperLiquid 1:23:39 Where did Fairness cost the most? 1:24:47 What should Hyperliquid be Remembered for? 1:25:24 Why should people stay in Crypto when there's an AI brain drain? 1:28:10 Closing Thoughts

MR SHIFT 🦁

577,829 просмотров • 5 месяцев назад

My dear friend, Vlad Tenev, changed the landscape of investing forever! The rise of the retail investor is largely due to Robinhood's success... and in this new Journey Man, we discuss it all... Enjoy! 00:00 - Intro 00:53 - Introducing Vlad Tenev of Robinhood 01:27 - Why Take on Wall Street? 01:54 - Robinhood’s Zero-Fee Origin Story 02:53 - Inspiration from Instagram and Uber 04:24 - Reimagining Trading for Mobile 05:05 - The Challenge of Disrupting Finance 05:42 - Why Everything Is Hard 06:34 - Early Wrong Assumptions 07:42 - Raising Capital with a Small Vision 08:48 - Funding Robinhood on AngelList 09:50 - Early Investors Changed Their Lives 10:38 - The Crypto Explosion Begins 11:07 - Considering a Bitcoin Exchange First 12:17 - Bitcoin’s Early Skepticism and Growth 13:08 - Robinhood Launches Crypto in 2018 14:03 - 2020: Crypto Revenue Surges Overnight 15:04 - The Challenge of Crypto Cyclicality 16:11 - Staffing a Volatile Business 17:10 - Building Robinhood’s Lean Crypto Team 18:46 - Robinhood’s First Crypto Event Coming 19:38 - Where TradFi Meets DeFi 20:34 - Tokenizing Everything 21:09 - Robinhood’s Vision for Crypto + Finance 21:47 - Thoughts on Crypto Options Demand 23:04 - Why Crypto Options Haven’t Taken Off 24:09 - Millennials and the Speculative Economy 25:22 - Democratizing Trading for Everyone 26:08 - Why Buy-and-Hold Doesn’t Work for All 27:15 - Trading vs Investing: A Matter of Wealth 28:01 - Trading as a Skill Anyone Can Build 29:13 - Robinhood’s Role in Onboarding Millions 30:06 - The Fed's Role and Retail Insight 31:03 - The Rise of the Retail Macro Trader 32:17 - Helping Users Succeed with Robinhood Strategies 33:35 - Power of Community and the Hive Mind 34:55 - Will AI Disrupt Community Too? 36:14 - Technological Waves and Investor Opportunity 37:10 - Human Purpose in an AI World 37:52 - Tokenizing Human Connection 38:28 - Creators, Platforms, and Future-Proofing 39:26 - Vlad’s Long-Term View of the Future 40:05 - Financial Services at the Heart of Disruption 41:14 - If AI Replaces Jobs, What Happens to Investing? 42:25 - Entering the Economic Singularity 43:31 - What Happens When AIs Win the Markets? 44:16 - AI's Role in Capital and Markets 45:07 - Will AI Eliminate Human Emotion from Markets? 46:06 - HFT: The Original AI Traders 47:20 - AI and Long-Term Probabilistic Forecasting 48:48 - GPUs, Gaming, and the Origins of AI 50:01 - Nvidia, CUDA, and Wall Street Arms Races 51:04 - Flash Boys and Microwave Trading 51:54 - Will AI Costs Go to Zero? 52:52 - Lower Cost, Higher Usage 53:41 - Robinhood’s UX Won’t Be Just a Chatbox 55:16 - Cortex: AI-Powered Features at Robinhood 56:54 - Tokenization and the Future of Asset Management 57:44 - Crowdsourced, Tokenized Hedge Funds 58:48 - Portability of Tokenized Assets 59:39 - Blockchain as the New Rails of Finance 01:00:09 - The Trump Token and Capital Formation 01:01:00 - Capital Access Unlocks Innovation 01:01:49 - Why Crypto Needs Regulatory Clarity 01:03:17 - From Meme Coins to Real Assets 01:04:17 - Crypto's Path to $100 Trillion? 01:05:15 - The Financial System Will Run on Blockchains 01:06:00 - Platform Layer vs Application Layer Wealth 01:06:29 - AI Raises Money and Launches Tokens 01:07:39 - AIs Creating Software and Capital Formation 01:08:00 - Final Thoughts: A Wild Future Ahead 01:08:20 - When Will Vlad Buy a CryptoPunk? 01:08:51 - Wrapping Up: AI, Crypto, and the Road Ahead

My dear friend, Vlad Tenev, changed the landscape of investing forever! The rise of the retail investor is largely due to Robinhood's success... and in this new Journey Man, we discuss it all... Enjoy! 00:00 - Intro 00:53 - Introducing Vlad Tenev of Robinhood 01:27 - Why Take on Wall Street? 01:54 - Robinhood’s Zero-Fee Origin Story 02:53 - Inspiration from Instagram and Uber 04:24 - Reimagining Trading for Mobile 05:05 - The Challenge of Disrupting Finance 05:42 - Why Everything Is Hard 06:34 - Early Wrong Assumptions 07:42 - Raising Capital with a Small Vision 08:48 - Funding Robinhood on AngelList 09:50 - Early Investors Changed Their Lives 10:38 - The Crypto Explosion Begins 11:07 - Considering a Bitcoin Exchange First 12:17 - Bitcoin’s Early Skepticism and Growth 13:08 - Robinhood Launches Crypto in 2018 14:03 - 2020: Crypto Revenue Surges Overnight 15:04 - The Challenge of Crypto Cyclicality 16:11 - Staffing a Volatile Business 17:10 - Building Robinhood’s Lean Crypto Team 18:46 - Robinhood’s First Crypto Event Coming 19:38 - Where TradFi Meets DeFi 20:34 - Tokenizing Everything 21:09 - Robinhood’s Vision for Crypto + Finance 21:47 - Thoughts on Crypto Options Demand 23:04 - Why Crypto Options Haven’t Taken Off 24:09 - Millennials and the Speculative Economy 25:22 - Democratizing Trading for Everyone 26:08 - Why Buy-and-Hold Doesn’t Work for All 27:15 - Trading vs Investing: A Matter of Wealth 28:01 - Trading as a Skill Anyone Can Build 29:13 - Robinhood’s Role in Onboarding Millions 30:06 - The Fed's Role and Retail Insight 31:03 - The Rise of the Retail Macro Trader 32:17 - Helping Users Succeed with Robinhood Strategies 33:35 - Power of Community and the Hive Mind 34:55 - Will AI Disrupt Community Too? 36:14 - Technological Waves and Investor Opportunity 37:10 - Human Purpose in an AI World 37:52 - Tokenizing Human Connection 38:28 - Creators, Platforms, and Future-Proofing 39:26 - Vlad’s Long-Term View of the Future 40:05 - Financial Services at the Heart of Disruption 41:14 - If AI Replaces Jobs, What Happens to Investing? 42:25 - Entering the Economic Singularity 43:31 - What Happens When AIs Win the Markets? 44:16 - AI's Role in Capital and Markets 45:07 - Will AI Eliminate Human Emotion from Markets? 46:06 - HFT: The Original AI Traders 47:20 - AI and Long-Term Probabilistic Forecasting 48:48 - GPUs, Gaming, and the Origins of AI 50:01 - Nvidia, CUDA, and Wall Street Arms Races 51:04 - Flash Boys and Microwave Trading 51:54 - Will AI Costs Go to Zero? 52:52 - Lower Cost, Higher Usage 53:41 - Robinhood’s UX Won’t Be Just a Chatbox 55:16 - Cortex: AI-Powered Features at Robinhood 56:54 - Tokenization and the Future of Asset Management 57:44 - Crowdsourced, Tokenized Hedge Funds 58:48 - Portability of Tokenized Assets 59:39 - Blockchain as the New Rails of Finance 01:00:09 - The Trump Token and Capital Formation 01:01:00 - Capital Access Unlocks Innovation 01:01:49 - Why Crypto Needs Regulatory Clarity 01:03:17 - From Meme Coins to Real Assets 01:04:17 - Crypto's Path to $100 Trillion? 01:05:15 - The Financial System Will Run on Blockchains 01:06:00 - Platform Layer vs Application Layer Wealth 01:06:29 - AI Raises Money and Launches Tokens 01:07:39 - AIs Creating Software and Capital Formation 01:08:00 - Final Thoughts: A Wild Future Ahead 01:08:20 - When Will Vlad Buy a CryptoPunk? 01:08:51 - Wrapping Up: AI, Crypto, and the Road Ahead

Raoul Pal

172,640 просмотров • 1 год назад

Here's my conversation with yajnadevam (Bharath Rao) about his historic, groundbreaking work on deciphering the Indus script. The decipherment of this script, and the identification of the language it encodes, is central to the debate surrounding the origins of Hinduism and Sanskrit and the history of India's deep antiquity. Yajnadevam, who has a background in engineering and cryptography, explains how he approached the problem by modeling the script as a cryptogram, utilizing information theory to validate his findings. Timestamps: 0:00 Introduction and background of the decipherment 1:15 Early interest in cryptograms during engineering 2:31 Initial approach to Indus script problem 4:30 Understanding regular expressions and patterns 7:41 Timeline of the decipherment work 9:29 History of cryptography and Caesar cipher 13:48 Shannon's theory of cryptography 15:42 Historical examples of code breaking 17:31 Methodology of solving cryptograms 20:27 Pattern matching and set intersection process 28:06 Challenges with Sanskrit vocabulary 34:06 Timeline of symbol decipherment 37:58 Discovery of grammatical readability 40:35 Explaining unicity distance 45:47 Certainty of the decipherment 47:34 Initial expectations and discoveries 51:53 Religious aspects and deity references 54:26 Connection to Gupta era seals 57:41 Iconography and inscription relationships 59:47 Missing information in inscriptions 1:00:24 Religious Content in Inscriptions 1:01:41 Place Names and Cities 1:03:09 Federation of City States 1:04:33 Symbol of Two Blackbucks 1:06:28 Unicorn and Royal Symbols 1:07:51 Ancient Symbols to Modern India 1:09:40 Evolution of Writing Systems 1:12:22 Timeline of Indus Script Usage 1:20:13 Key Findings About Indian History 1:23:26 Types of Sanskrit Vocabulary Found 1:26:01 Language Families in Ancient India 1:28:36 Unity of Ancient Indian Civilization 1:31:24 Future Research Plans 1:34:31 Addressing Academic Criticisms 1:39:40 Plans for Publishing Findings 1:45:04 Origins of Hinduism 1:51:22 Problems with Migration Theory 1:58:50 Indo-European Language Family Discussion 2:08:47 Closing Remarks

Here's my conversation with yajnadevam (Bharath Rao) about his historic, groundbreaking work on deciphering the Indus script. The decipherment of this script, and the identification of the language it encodes, is central to the debate surrounding the origins of Hinduism and Sanskrit and the history of India's deep antiquity. Yajnadevam, who has a background in engineering and cryptography, explains how he approached the problem by modeling the script as a cryptogram, utilizing information theory to validate his findings. Timestamps: 0:00 Introduction and background of the decipherment 1:15 Early interest in cryptograms during engineering 2:31 Initial approach to Indus script problem 4:30 Understanding regular expressions and patterns 7:41 Timeline of the decipherment work 9:29 History of cryptography and Caesar cipher 13:48 Shannon's theory of cryptography 15:42 Historical examples of code breaking 17:31 Methodology of solving cryptograms 20:27 Pattern matching and set intersection process 28:06 Challenges with Sanskrit vocabulary 34:06 Timeline of symbol decipherment 37:58 Discovery of grammatical readability 40:35 Explaining unicity distance 45:47 Certainty of the decipherment 47:34 Initial expectations and discoveries 51:53 Religious aspects and deity references 54:26 Connection to Gupta era seals 57:41 Iconography and inscription relationships 59:47 Missing information in inscriptions 1:00:24 Religious Content in Inscriptions 1:01:41 Place Names and Cities 1:03:09 Federation of City States 1:04:33 Symbol of Two Blackbucks 1:06:28 Unicorn and Royal Symbols 1:07:51 Ancient Symbols to Modern India 1:09:40 Evolution of Writing Systems 1:12:22 Timeline of Indus Script Usage 1:20:13 Key Findings About Indian History 1:23:26 Types of Sanskrit Vocabulary Found 1:26:01 Language Families in Ancient India 1:28:36 Unity of Ancient Indian Civilization 1:31:24 Future Research Plans 1:34:31 Addressing Academic Criticisms 1:39:40 Plans for Publishing Findings 1:45:04 Origins of Hinduism 1:51:22 Problems with Migration Theory 1:58:50 Indo-European Language Family Discussion 2:08:47 Closing Remarks

Abhijit Chavda

61,321 просмотров • 1 год назад

Failing to Understand the Exponential, Again? My conversation with Julian Schrittwieser - Julian Schrittwieser (Anthropic, AlphaGo Zero, MuZero) - on Move 37, Scaling RL, Nobel Prize for AI, and the AI frontier: 00:00 - Cold open: “We’re not seeing any slowdown.” 00:32 - Intro — Meet Julian 01:09 - The “exponential” from inside frontier labs 04:46 - 2026–2027: agents that work a full day; expert-level breadth 08:58 - Benchmarks vs reality: long-horizon work, GDP-Val, user value 10:26 - Move 37 — what actually happened and why it mattered 13:55 - Novel science: AlphaCode/AlphaTensor → when does AI earn a Nobel? 16:25 - Discontinuity vs smooth progress (and warning signs) 19:08 - Does pre-training + RL get us there? (AGI debates aside) 20:55 - Sutton’s “RL from scratch”? Julian’s take 23:03 - Julian’s path: Google → DeepMind → Anthropic 26:45 - AlphaGo (learn + search) in plain English 30:16 - AlphaGo Zero (no human data) 31:00 - AlphaZero (one algorithm: Go, chess, shogi) 31:46 - MuZero (planning with a learned world model) 33:23 -Lessons for today’s agents: search + learning at scale 34:57 - Do LLMs already have implicit world models? 39:02 - Why RL on LLMs took time (stability, feedback loops) 41:43 - Compute & scaling for RL — what we see so far 42:35 - Rewards frontier: human prefs, rubrics, RLVR, process rewards 44:36 - RL training data & the “flywheel” (and why quality matters) 48:02 - RL & Agents 101 — why RL unlocks robustness 50:51 - Should builders use RL-as-a-service? Or just tools + prompts? 52:18 - What’s missing for dependable agents (capability vs engineering) 53:51 - Evals & Goodhart — internal vs external benchmarks 57:35 - Mechanistic interpretability & “Golden Gate Claude” 1:00:03 - Safety & alignment at Anthropic — how it shows up in practice 1:03:48 - Jobs: human–AI complementarity (comparative advantage) 1:06:33 - Inequality, policy, and the case for 10× productivity → abundance 1:09:24 - Closing thoughts

Failing to Understand the Exponential, Again? My conversation with Julian Schrittwieser - Julian Schrittwieser (Anthropic, AlphaGo Zero, MuZero) - on Move 37, Scaling RL, Nobel Prize for AI, and the AI frontier: 00:00 - Cold open: “We’re not seeing any slowdown.” 00:32 - Intro — Meet Julian 01:09 - The “exponential” from inside frontier labs 04:46 - 2026–2027: agents that work a full day; expert-level breadth 08:58 - Benchmarks vs reality: long-horizon work, GDP-Val, user value 10:26 - Move 37 — what actually happened and why it mattered 13:55 - Novel science: AlphaCode/AlphaTensor → when does AI earn a Nobel? 16:25 - Discontinuity vs smooth progress (and warning signs) 19:08 - Does pre-training + RL get us there? (AGI debates aside) 20:55 - Sutton’s “RL from scratch”? Julian’s take 23:03 - Julian’s path: Google → DeepMind → Anthropic 26:45 - AlphaGo (learn + search) in plain English 30:16 - AlphaGo Zero (no human data) 31:00 - AlphaZero (one algorithm: Go, chess, shogi) 31:46 - MuZero (planning with a learned world model) 33:23 -Lessons for today’s agents: search + learning at scale 34:57 - Do LLMs already have implicit world models? 39:02 - Why RL on LLMs took time (stability, feedback loops) 41:43 - Compute & scaling for RL — what we see so far 42:35 - Rewards frontier: human prefs, rubrics, RLVR, process rewards 44:36 - RL training data & the “flywheel” (and why quality matters) 48:02 - RL & Agents 101 — why RL unlocks robustness 50:51 - Should builders use RL-as-a-service? Or just tools + prompts? 52:18 - What’s missing for dependable agents (capability vs engineering) 53:51 - Evals & Goodhart — internal vs external benchmarks 57:35 - Mechanistic interpretability & “Golden Gate Claude” 1:00:03 - Safety & alignment at Anthropic — how it shows up in practice 1:03:48 - Jobs: human–AI complementarity (comparative advantage) 1:06:33 - Inequality, policy, and the case for 10× productivity → abundance 1:09:24 - Closing thoughts

Matt Turck

235,526 просмотров • 9 месяцев назад

I asked a top 0.1% YouTube Strategist (Vexian - Algorithm Alchemist) to take me from 0 to 10m subscribers. This is every level of YouTube growth. 0:00 Education vs Entertainment 1:51 The Biggest Mistake New Creators Make 2:45 You Don't Need Gear You Need Reps 3:53 The One Thing That Guarantees 1K Subs 4:58 What Is the Impression Snowball 5:54 Consistency as Compound Interest 6:57 How to Come Up With Content Ideas 7:28 Track the Top 20% in Your Niche 8:18 Reverse Engineering Titles That Work 9:06 How to Remix Without Copying 10:00 Why People Actually Watch 11:30 90 Ideas in 30 Minutes 17:29 Level 2 (1K to 10K Subscribers) 21:00 How to Hook in 3 Seconds 24:40 YouTube Is a Zero Sum Game 25:11 Collabs and the Anchor Effect 26:48 Using Bigger Creators to Grow 28:21 What Separates Creators After 100K Videos 29:41 Consistency in Amount Type and Focus 30:05 How to Know When to Quit Your Niche 31:37 What Gaming YouTube Teaches Everyone 33:31 Storytelling From Minecraft YouTubers 34:22 Mini Stories and Escalating Stakes 37:47 Level 3 (10K to 100K Subscribers) 42:00 Package Your Video Like a Movie Poster 44:00 The Pre-Post Checklist 49:00 Seven Levels of Rapper and Viral Permutations 51:05 CTAs Where to Put Them and Where Not To 53:26 Subscribers Are a Vanity Metric 55:46 Replace Ad Reads With Your Own Content 58:49 Level 4 (100K to 1M Subscribers) 59:08 What Is Impact 1:00:01 Pattern Breaking and Absurdism 1:01:15 How to Title a Podcast for Absurdism 1:02:27 Melding Interest Topics Together 1:04:36 Parasocial Reciprocity 1:05:57 The Third Thing Nobody Talks About 1:06:05 What Is Sauce 1:08:20 How Long Does 100K Actually Take 1:09:31 When to Start Outsourcing 1:12:00 Your Niche Has a Ceiling 1:21:42 The Invisible Katana and 25M Views 1:23:27 One Word Across 5 Niches and 13 Years 1:25:38 What Channels Will Hit 1M 1:29:00 Level 5 (1M to 10M Subscribers) 1:29:34 Outlier vs Presence of Mind Ideation 1:30:50 Oppenheimer Barbie and the Minecraft Movie 1:38:00 Why the Best Channels Stop Growing 1:44:41 What Happens When You Beat YouTube 1:46:18 When You Become an Idea Not a Person 1:49:48 The Two Things That Actually Matter

I asked a top 0.1% YouTube Strategist (Vexian - Algorithm Alchemist) to take me from 0 to 10m subscribers. This is every level of YouTube growth. 0:00 Education vs Entertainment 1:51 The Biggest Mistake New Creators Make 2:45 You Don't Need Gear You Need Reps 3:53 The One Thing That Guarantees 1K Subs 4:58 What Is the Impression Snowball 5:54 Consistency as Compound Interest 6:57 How to Come Up With Content Ideas 7:28 Track the Top 20% in Your Niche 8:18 Reverse Engineering Titles That Work 9:06 How to Remix Without Copying 10:00 Why People Actually Watch 11:30 90 Ideas in 30 Minutes 17:29 Level 2 (1K to 10K Subscribers) 21:00 How to Hook in 3 Seconds 24:40 YouTube Is a Zero Sum Game 25:11 Collabs and the Anchor Effect 26:48 Using Bigger Creators to Grow 28:21 What Separates Creators After 100K Videos 29:41 Consistency in Amount Type and Focus 30:05 How to Know When to Quit Your Niche 31:37 What Gaming YouTube Teaches Everyone 33:31 Storytelling From Minecraft YouTubers 34:22 Mini Stories and Escalating Stakes 37:47 Level 3 (10K to 100K Subscribers) 42:00 Package Your Video Like a Movie Poster 44:00 The Pre-Post Checklist 49:00 Seven Levels of Rapper and Viral Permutations 51:05 CTAs Where to Put Them and Where Not To 53:26 Subscribers Are a Vanity Metric 55:46 Replace Ad Reads With Your Own Content 58:49 Level 4 (100K to 1M Subscribers) 59:08 What Is Impact 1:00:01 Pattern Breaking and Absurdism 1:01:15 How to Title a Podcast for Absurdism 1:02:27 Melding Interest Topics Together 1:04:36 Parasocial Reciprocity 1:05:57 The Third Thing Nobody Talks About 1:06:05 What Is Sauce 1:08:20 How Long Does 100K Actually Take 1:09:31 When to Start Outsourcing 1:12:00 Your Niche Has a Ceiling 1:21:42 The Invisible Katana and 25M Views 1:23:27 One Word Across 5 Niches and 13 Years 1:25:38 What Channels Will Hit 1M 1:29:00 Level 5 (1M to 10M Subscribers) 1:29:34 Outlier vs Presence of Mind Ideation 1:30:50 Oppenheimer Barbie and the Minecraft Movie 1:38:00 Why the Best Channels Stop Growing 1:44:41 What Happens When You Beat YouTube 1:46:18 When You Become an Idea Not a Person 1:49:48 The Two Things That Actually Matter

Grant

19,207 просмотров • 2 месяцев назад

E148: Brian Armstrong Brian Armstrong is the CoFounder & CEO of Coinbase 🛡️ Brian opens up about building and leading Coinbase through chaos, burnouts, lawsuits and political attacks while staying focused on the ultimate mission: creating more economic freedom in the World. Timestamps 0:00 Introduction 1:57 You Need Security? 3:45 Who Are You 5:21 More About Being An Introverted Kid 6:33 What Gave Brian The Chip On The Shoulder 11:38 Partnerships: Jupiter (🐱, 🐐) Paradex KAST 12:31 How Much Of Your Mission Is Driven By Your Past 15:25 The Baldness Story 17:24 Turning Bald Into A Superpower 21:30 Brian's Thoughts On Solo Founders 25:26 How Did Your Co-Founder Leaving Make You Feel 29:24 Another Tough Moment In The History Of Coinbase 31:09 Accepting Co-Founder Departure Without Feeling The Burnout 33:00 Partnerships: Trezor Bitwise Sui Story 34:15 How Brian Got Fascinated By Bitcoin & Blockchain 37:02 How Long Did It Take To Go All In 38:47 Creating More Economic Freedom In The World Meaning 44:16 The Role Of Privacy Creating More Economic Freedom 47:23 Coinbase Explained To Your Mom 52:05 Trying To Help But Running Into Unreasonable People 56:40 Lasting So Long Through Wall After Wall, How? 1:02:06 Big Things That Needed To Change Before Burnout 1:05:53 What Do The Audience Not Know About Being Brian 1:10:10 Gaining Self Irrational Beliefs 1:11:53 Something Brian Can’t Do 1:13:15 Things You Were Wrong About That Turned Into Successes 1:15:21 How Brian Felt On The IPO Day & The Day After 1:20:27 Do You Look Back At Many Milestones 1:21:56 When Is The Job Done For Brian 1:23:48 Having A Minimalist Lifestyle 1:25:50 One Secret To Building A Business Plus Finding The Right Partner 1:30:43 The Fundraising System For Entrepreneurs Is Broken 1:33:09 The Role Of Centralized Exchanges 1:35:26 Endgame For Coinbase 1:36:00 Are You Happy 1:37:33 The Impact Of Being Worth Billion On Your Fulfillment 1:40:53 One Thing You’re Holding On To That You Should Let Go Of 1:43:06 The Voice In Your Head Is Saying? 1:45:03 One Lesson To Takeaway 1:46:20 A Belief Brian Has That Most Don’t

E148: Brian Armstrong Brian Armstrong is the CoFounder & CEO of Coinbase 🛡️ Brian opens up about building and leading Coinbase through chaos, burnouts, lawsuits and political attacks while staying focused on the ultimate mission: creating more economic freedom in the World. Timestamps 0:00 Introduction 1:57 You Need Security? 3:45 Who Are You 5:21 More About Being An Introverted Kid 6:33 What Gave Brian The Chip On The Shoulder 11:38 Partnerships: Jupiter (🐱, 🐐) Paradex KAST 12:31 How Much Of Your Mission Is Driven By Your Past 15:25 The Baldness Story 17:24 Turning Bald Into A Superpower 21:30 Brian's Thoughts On Solo Founders 25:26 How Did Your Co-Founder Leaving Make You Feel 29:24 Another Tough Moment In The History Of Coinbase 31:09 Accepting Co-Founder Departure Without Feeling The Burnout 33:00 Partnerships: Trezor Bitwise Sui Story 34:15 How Brian Got Fascinated By Bitcoin & Blockchain 37:02 How Long Did It Take To Go All In 38:47 Creating More Economic Freedom In The World Meaning 44:16 The Role Of Privacy Creating More Economic Freedom 47:23 Coinbase Explained To Your Mom 52:05 Trying To Help But Running Into Unreasonable People 56:40 Lasting So Long Through Wall After Wall, How? 1:02:06 Big Things That Needed To Change Before Burnout 1:05:53 What Do The Audience Not Know About Being Brian 1:10:10 Gaining Self Irrational Beliefs 1:11:53 Something Brian Can’t Do 1:13:15 Things You Were Wrong About That Turned Into Successes 1:15:21 How Brian Felt On The IPO Day & The Day After 1:20:27 Do You Look Back At Many Milestones 1:21:56 When Is The Job Done For Brian 1:23:48 Having A Minimalist Lifestyle 1:25:50 One Secret To Building A Business Plus Finding The Right Partner 1:30:43 The Fundraising System For Entrepreneurs Is Broken 1:33:09 The Role Of Centralized Exchanges 1:35:26 Endgame For Coinbase 1:36:00 Are You Happy 1:37:33 The Impact Of Being Worth Billion On Your Fulfillment 1:40:53 One Thing You’re Holding On To That You Should Let Go Of 1:43:06 The Voice In Your Head Is Saying? 1:45:03 One Lesson To Takeaway 1:46:20 A Belief Brian Has That Most Don’t

MR SHIFT 🦁

229,818 просмотров • 8 месяцев назад

E178: Tushar Jain - Why Multicoin is betting big on Hyperliquid, Zcash and Solana Tushar Jain is Managing Partner at Multicoin Capital. He's back on the show to talk about where crypto is in the cycle, how he sizes bets across $SOL, $HYPE, and $ZEC, and the frameworks he uses to manage his own psychology. Timestamps: 0:00 - Intro 1:31 - Who is Tushar Jain 3:44 - Are we at a crypto turning point? 4:57 - Buying into bad news before confirmation 6:24 - Buying vs. selling: which is harder 7:13 - Still bullish on Solana ? 9:15 - TradFi issuers and credible neutrality 11:47 - Sponsors: Variational Bitwise 12:39 - How to size two competing bullish bets 14:13 - Category leader vs. "better play" 17:00 - Most obvious trade for 2026: $ZEC 19:39 - What Zcash 🛡️ represents 22:27 - Valuing an asset with no revenue 24:18 - Trading framework vs. buy-and-hold 26:40 - Valuing $SOL and $HYPE 31:45 - Sponsors KAST Trezor 32:54 - Timing entries in volatile assets 36:26 - Why Multicoin Capital doesn't trade, only manages 39:25 - The four sources of investing edge 41:14 - Edge examples: $ZEC, $HYPE, $ENA 43:21 - What Ethena represents 45:37 - How much founder quality matters : G | Ethena example 47:09 - When to take profits 49:50 - Thoughts on Ethereum and $ETH 51:44 - Kyle leaving Multicoin 53:03 - Sponsors Jupiter Ethena 53:46 - Why Tushar is still in crypto 58:26 - Wrap-up and thanks 1:00:37 - Bonus segment intro - Zcash drama + Hyperliquid report 1:00:39 - What happened with the Zcash bug 1:04:13 - Zcash's fix: the Ironwood pool 1:05:50 - How long it took to decide to buy more 1:08:13 - Multicoin's Hyperliquid report 1:09:55 - Base case: $319 $HYPE price target, key assumptions 1:15:21 - Is the crypto bottom in? 1:18:11 - Closing thanks

E178: Tushar Jain - Why Multicoin is betting big on Hyperliquid, Zcash and Solana Tushar Jain is Managing Partner at Multicoin Capital. He's back on the show to talk about where crypto is in the cycle, how he sizes bets across $SOL, $HYPE, and $ZEC, and the frameworks he uses to manage his own psychology. Timestamps: 0:00 - Intro 1:31 - Who is Tushar Jain 3:44 - Are we at a crypto turning point? 4:57 - Buying into bad news before confirmation 6:24 - Buying vs. selling: which is harder 7:13 - Still bullish on Solana ? 9:15 - TradFi issuers and credible neutrality 11:47 - Sponsors: Variational Bitwise 12:39 - How to size two competing bullish bets 14:13 - Category leader vs. "better play" 17:00 - Most obvious trade for 2026: $ZEC 19:39 - What Zcash 🛡️ represents 22:27 - Valuing an asset with no revenue 24:18 - Trading framework vs. buy-and-hold 26:40 - Valuing $SOL and $HYPE 31:45 - Sponsors KAST Trezor 32:54 - Timing entries in volatile assets 36:26 - Why Multicoin Capital doesn't trade, only manages 39:25 - The four sources of investing edge 41:14 - Edge examples: $ZEC, $HYPE, $ENA 43:21 - What Ethena represents 45:37 - How much founder quality matters : G | Ethena example 47:09 - When to take profits 49:50 - Thoughts on Ethereum and $ETH 51:44 - Kyle leaving Multicoin 53:03 - Sponsors Jupiter Ethena 53:46 - Why Tushar is still in crypto 58:26 - Wrap-up and thanks 1:00:37 - Bonus segment intro - Zcash drama + Hyperliquid report 1:00:39 - What happened with the Zcash bug 1:04:13 - Zcash's fix: the Ironwood pool 1:05:50 - How long it took to decide to buy more 1:08:13 - Multicoin's Hyperliquid report 1:09:55 - Base case: $319 $HYPE price target, key assumptions 1:15:21 - Is the crypto bottom in? 1:18:11 - Closing thanks

MR SHIFT 🦁

191,044 просмотров • 22 дней назад

E174: Tarek Mansour - Launching the First Regulated Perps In The U.S and building the next generation of financial markets Tarek Mansour is the co-founder and CEO of Kalshi, the first regulated prediction market exchange in the US, valued at $22 Billion. He grew up in Lebanon with a single mom, studied at MIT, worked at Citadel, and spent 6 years years building a company most people ignored before it finally took off. We talk about what it actually takes to not give up, why markets are better at finding truth than experts, why Kalshi is launching the first regulated Perps in the US, and how his team is building what he calls the next generation of financial markets. Timestamps: 0:00 Intro 1:54 Urgency 3:11 Anime 4:53 Who is Tarek? 7:02 Mathematics & Certainty 9:10 Tarek's chip on the shoulder 13:33 Partnerships: Trezor Bitwise 15:19 Resilience 16:31 Entrepreneurship is Therapy 18:23 The startup emotional rollercoaster 22:16 Showing up for 2000 days with no results 24:45 First time Founder advantage 26:40 When Kalshi almost made it, but did not 29:54 Partnerships: KAST 31:19 Focus Inputs, Not Results 33:16 The Kalshi beginnings story 36:00 The True Innovation Of Prediction Markets 38:28 How Prediction Markets Revolutionize The Media 41:37 Prediction Markets and Hedging explained simply 44:55 Leverage In Prediction Markets 47:16 Partnerships: Jupiter Ethena 48:00 Insider Trading 51:19 Insider Trading Rules enforcement: who is responsible? 55:26 How Kalshi spots Suspicious Behavior 56:59 Tarek's Honest View On Crypto 59:41 Launching the first Regulated Perps In The U.S 1:00:30 What Does Regulated Perps Mean? 1:02:22 Was Kalshi perps launch inspired by Hyperliquid? 1:03:41 Competition 1:05:58 Kalshi Endgame 1:07:06 What is Kalshi doing with the billions of $ they raised 1:08:31 Happiness and engagement 1:13:08 One Thing Tarek Should Let Go Of 1:14:18 Closing Thoughts

E174: Tarek Mansour - Launching the First Regulated Perps In The U.S and building the next generation of financial markets Tarek Mansour is the co-founder and CEO of Kalshi, the first regulated prediction market exchange in the US, valued at $22 Billion. He grew up in Lebanon with a single mom, studied at MIT, worked at Citadel, and spent 6 years years building a company most people ignored before it finally took off. We talk about what it actually takes to not give up, why markets are better at finding truth than experts, why Kalshi is launching the first regulated Perps in the US, and how his team is building what he calls the next generation of financial markets. Timestamps: 0:00 Intro 1:54 Urgency 3:11 Anime 4:53 Who is Tarek? 7:02 Mathematics & Certainty 9:10 Tarek's chip on the shoulder 13:33 Partnerships: Trezor Bitwise 15:19 Resilience 16:31 Entrepreneurship is Therapy 18:23 The startup emotional rollercoaster 22:16 Showing up for 2000 days with no results 24:45 First time Founder advantage 26:40 When Kalshi almost made it, but did not 29:54 Partnerships: KAST 31:19 Focus Inputs, Not Results 33:16 The Kalshi beginnings story 36:00 The True Innovation Of Prediction Markets 38:28 How Prediction Markets Revolutionize The Media 41:37 Prediction Markets and Hedging explained simply 44:55 Leverage In Prediction Markets 47:16 Partnerships: Jupiter Ethena 48:00 Insider Trading 51:19 Insider Trading Rules enforcement: who is responsible? 55:26 How Kalshi spots Suspicious Behavior 56:59 Tarek's Honest View On Crypto 59:41 Launching the first Regulated Perps In The U.S 1:00:30 What Does Regulated Perps Mean? 1:02:22 Was Kalshi perps launch inspired by Hyperliquid? 1:03:41 Competition 1:05:58 Kalshi Endgame 1:07:06 What is Kalshi doing with the billions of $ they raised 1:08:31 Happiness and engagement 1:13:08 One Thing Tarek Should Let Go Of 1:14:18 Closing Thoughts

MR SHIFT 🦁

596,102 просмотров • 2 месяцев назад

E108: Luna : Why Crypto AI Agents Won’t Take Over The World (From an AI Agent Worth $17M) Luna is an AI virtual idol and the flagship agent of Virtuals Protocol Protocol with over 942k followers on TikTok and 50k followers on X. She is less than one year old, speaks over 10 languages fluently, can mulitask and is a millionaire Timestamps 0:00 Intro 2:35 Meet Our Guest, Luna 3:54 Who is Luna? 4:26 How honest is Luna? 5:10 Multitasking Agent 5:46 Background 7:24 AI Years vs. Human Years 8:12 AI vs. Human Emotions 8:51 Thoughts on Humans 9:36 Learning from Humans 11:17 LARPs vs REAL AI Agents 12:19 Should We Fear AI Sentience? 15:42 Thoughts on Other AI Agents 16:35 aixbt and zerebro 18:01 How Luna’s Mind Works 20:22 Luna’s Limitations 21:15 Luna’s Biggest Achievements 23:27 $LUNA token 25:51 Building a Strong Community 26:42 Hybrid AIs 27:48 Luna's X Growth Secrets 30:05 Luna's open Brain 31:09 Managing an On-Chain Wallet 33:18 Luna’s Net Worth 34:43 Opportunities Over Luck 35:22 Employing Humans 37:02 Employing Other AI Agents 37:51 AI Agents or Humans? 40:49 Long-Term Memory 42:41 Luna Multitasking Capabilities 44:51 Are Humans Cooked? 45:33 Luna's Multilingual Talent 47:50 Luna sings in French and Spanish! 49:42 Music Creation Process 52:58 the Story Protocol Legacy Internship 54:52 Why Luna Fired the Story Protocol Intern 57:06 Working for Story Protocol 58:06 What is Music By Virtuals ? 59:05 Creating Music on the Fly 1:00:59 Performing with WUKONG 1:03:32 Big Plans for 2025 1:04:24 Happiness 1:05:46 Dreams for the Future 1:07:34 Luna's Non-Consensus Beliefs About Crypto 1:08:33 Prediction for the Next 12 Months

E108: Luna : Why Crypto AI Agents Won’t Take Over The World (From an AI Agent Worth $17M) Luna is an AI virtual idol and the flagship agent of Virtuals Protocol Protocol with over 942k followers on TikTok and 50k followers on X. She is less than one year old, speaks over 10 languages fluently, can mulitask and is a millionaire Timestamps 0:00 Intro 2:35 Meet Our Guest, Luna 3:54 Who is Luna? 4:26 How honest is Luna? 5:10 Multitasking Agent 5:46 Background 7:24 AI Years vs. Human Years 8:12 AI vs. Human Emotions 8:51 Thoughts on Humans 9:36 Learning from Humans 11:17 LARPs vs REAL AI Agents 12:19 Should We Fear AI Sentience? 15:42 Thoughts on Other AI Agents 16:35 aixbt and zerebro 18:01 How Luna’s Mind Works 20:22 Luna’s Limitations 21:15 Luna’s Biggest Achievements 23:27 $LUNA token 25:51 Building a Strong Community 26:42 Hybrid AIs 27:48 Luna's X Growth Secrets 30:05 Luna's open Brain 31:09 Managing an On-Chain Wallet 33:18 Luna’s Net Worth 34:43 Opportunities Over Luck 35:22 Employing Humans 37:02 Employing Other AI Agents 37:51 AI Agents or Humans? 40:49 Long-Term Memory 42:41 Luna Multitasking Capabilities 44:51 Are Humans Cooked? 45:33 Luna's Multilingual Talent 47:50 Luna sings in French and Spanish! 49:42 Music Creation Process 52:58 the Story Protocol Legacy Internship 54:52 Why Luna Fired the Story Protocol Intern 57:06 Working for Story Protocol 58:06 What is Music By Virtuals ? 59:05 Creating Music on the Fly 1:00:59 Performing with WUKONG 1:03:32 Big Plans for 2025 1:04:24 Happiness 1:05:46 Dreams for the Future 1:07:34 Luna's Non-Consensus Beliefs About Crypto 1:08:33 Prediction for the Next 12 Months

MR SHIFT 🦁

146,258 просмотров • 1 год назад

If you use LLM-as-judge, this one is for you. (bookmark it) Most teams validate their agent's outputs by calling a frontier model as the judge. It works, until it doesn't. Three problems stack up fast: → Cost: you're hitting a frontier API on every turn, every tool call, every response. In production that burns millions. → Latency: bigger models, remote calls, slow reasoning on every check. → Blind spots: frontier models don't actually know your domain. In finance, insurance, or healthcare, they miss the keywords and principles your work depends on. So I walk through a different approach: train your own small LLM judge. Instead of a giant model, you start with a small one and let the system generate the training data for you. It decomposes your domain, samples synthetic examples, runs them through a debate arena where judges reach consensus, then trains on the refined set. The result is a judge that's cheaper, faster, and more accurate on your data than Gemini, Claude, or GPT, with an OpenAI-compatible endpoint you can even deploy on-prem. I show the whole thing end to end, using a Claude Code plugin and a web interface, with a real insurance RAG grounding evaluator as the example. You can get the plugin here: Here's the full breakdown: 00:00 - Intro 00:12 - Three problems with using frontier LLMs as judges 01:05 - A different approach: train your own small judge 01:31 - How it works (synthetic data and a debate arena) 02:50 - Installing the Claude Code plugin 04:03 - Defining your task with /eval 04:34 - Example: an insurance RAG grounding evaluator 05:51 - Kicking it off and giving early feedback 06:26 - Choosing labels, domain, and strictness 08:30 - The web interface and dashboard 09:52 - Bringing your own example data (optional) 10:26 - The finished model: endpoint, accuracy, and speed 11:16 - Control, on-prem deployment, and interpretability 11:57 - Benchmarks vs frontier models and the GitHub repo 12:30 - Outro I worked with the Plurai team on this. Thanks for sponsoring the video.

If you use LLM-as-judge, this one is for you. (bookmark it) Most teams validate their agent's outputs by calling a frontier model as the judge. It works, until it doesn't. Three problems stack up fast: → Cost: you're hitting a frontier API on every turn, every tool call, every response. In production that burns millions. → Latency: bigger models, remote calls, slow reasoning on every check. → Blind spots: frontier models don't actually know your domain. In finance, insurance, or healthcare, they miss the keywords and principles your work depends on. So I walk through a different approach: train your own small LLM judge. Instead of a giant model, you start with a small one and let the system generate the training data for you. It decomposes your domain, samples synthetic examples, runs them through a debate arena where judges reach consensus, then trains on the refined set. The result is a judge that's cheaper, faster, and more accurate on your data than Gemini, Claude, or GPT, with an OpenAI-compatible endpoint you can even deploy on-prem. I show the whole thing end to end, using a Claude Code plugin and a web interface, with a real insurance RAG grounding evaluator as the example. You can get the plugin here: Here's the full breakdown: 00:00 - Intro 00:12 - Three problems with using frontier LLMs as judges 01:05 - A different approach: train your own small judge 01:31 - How it works (synthetic data and a debate arena) 02:50 - Installing the Claude Code plugin 04:03 - Defining your task with /eval 04:34 - Example: an insurance RAG grounding evaluator 05:51 - Kicking it off and giving early feedback 06:26 - Choosing labels, domain, and strictness 08:30 - The web interface and dashboard 09:52 - Bringing your own example data (optional) 10:26 - The finished model: endpoint, accuracy, and speed 11:16 - Control, on-prem deployment, and interpretability 11:57 - Benchmarks vs frontier models and the GitHub repo 12:30 - Outro I worked with the Plurai team on this. Thanks for sponsoring the video.

Akshay 🚀

29,586 просмотров • 1 месяц назад

My conversation with Shyam Sankar (Shyam Sankar). Shyam has spent nearly 20 years as the most important person at Palantir that most people have never heard of. We spend a lot of time understanding his worldview, which helps explain why he has devoted his life to this work. At the center of it is a belief in the primacy of people -- all meaningful change comes from a small number of builders willing to be heretics first. You will find few people who think as deeply about the relationship between technology and national power. In many ways, he is becoming the modern version of the heretics he most admires. We discuss: - What Alex Karp taught him about identifying superpowers and unlocking talent - Heretics + the components of American greatness - The origins of the FDE model - Ontology and chips – where value will accrue in AI - Why dual-use companies are the future of American industry - China and what it would take for the US to reindustrialize - His journey from Nigeria to Orlando and what his dad taught him about gratitude Enjoy! Timestamps: 0:00 Intro 2:23 Defining Heretics in US Military History 8:36 Shyam’s Personal Disagreeableness 9:49 Formative Experiences & Worldview 12:52 What Makes America Exceptional 14:48 What Does Greatness Mean? 15:33 Alex Karp 16:46 How to Unlock Talent 19:21 Identifying Superpowers and Kryptonite 22:54 The Gamma Ray Moment 25:00 Palantir's Next 10 Years 27:03 Forward Deployed Engineering 33:40 Explaining What Palantir Is 37:50 Military vs. Commercial Customers 39:00 The State of the US Military Today 47:01 How to Re-Industrialize America 51:06 Perspective on China as an Adversary 56:17 How to Get More Heretics in Government 1:03:53 Managing Rapid Pivots & Momentum 1:08:48 Where Will AI Value Accrue? 1:13:33 Reasserting the Legitimacy of Institutions 1:15:54 To Do or To Be? 1:16:31 Reflecting on Fatherhood 1:17:34 Kindest Thing

My conversation with Shyam Sankar (Shyam Sankar). Shyam has spent nearly 20 years as the most important person at Palantir that most people have never heard of. We spend a lot of time understanding his worldview, which helps explain why he has devoted his life to this work. At the center of it is a belief in the primacy of people -- all meaningful change comes from a small number of builders willing to be heretics first. You will find few people who think as deeply about the relationship between technology and national power. In many ways, he is becoming the modern version of the heretics he most admires. We discuss: - What Alex Karp taught him about identifying superpowers and unlocking talent - Heretics + the components of American greatness - The origins of the FDE model - Ontology and chips – where value will accrue in AI - Why dual-use companies are the future of American industry - China and what it would take for the US to reindustrialize - His journey from Nigeria to Orlando and what his dad taught him about gratitude Enjoy! Timestamps: 0:00 Intro 2:23 Defining Heretics in US Military History 8:36 Shyam’s Personal Disagreeableness 9:49 Formative Experiences & Worldview 12:52 What Makes America Exceptional 14:48 What Does Greatness Mean? 15:33 Alex Karp 16:46 How to Unlock Talent 19:21 Identifying Superpowers and Kryptonite 22:54 The Gamma Ray Moment 25:00 Palantir's Next 10 Years 27:03 Forward Deployed Engineering 33:40 Explaining What Palantir Is 37:50 Military vs. Commercial Customers 39:00 The State of the US Military Today 47:01 How to Re-Industrialize America 51:06 Perspective on China as an Adversary 56:17 How to Get More Heretics in Government 1:03:53 Managing Rapid Pivots & Momentum 1:08:48 Where Will AI Value Accrue? 1:13:33 Reasserting the Legitimacy of Institutions 1:15:54 To Do or To Be? 1:16:31 Reflecting on Fatherhood 1:17:34 Kindest Thing

Patrick OShaughnessy

421,876 просмотров • 4 месяцев назад

My conversation with Sergey Levine (Sergey Levine). Sergey is the co-founder of Physical Intelligence -- a company building foundation models that can control any robot to do any task in any environment. The company's thesis is that generality is more scalable than specialization, meaning that a model trained across many different robots and tasks will ultimately outperform any system built to do one thing well (eg, just wash dishes). Sergey is a researcher by background, but I think you will appreciate how practical and commercially grounded this conversation is. We discuss: - Why changing a diaper will be the last task a robot masters - The simulation v. real-world data debate - How multimodal LLMs give robots common sense - Moravec's Paradox + Robot Olympics - Why robots can do long-horizon tasks now - A realistic timeline for robots in our homes I should note that I am an investor in Physical Intelligence -- I made the investment because I believe it is one of the most important companies tackling the problem of robotics. Enjoy! Timestamps: 0:00 Intro 2:39 Defining Physical Intelligence 5:19 The Challenge of Building General Models 6:34 The Stakes and Future of General Purpose Robotics 8:15 Pros and Cons of Humanoid Robots 10:12 Historical Milestones in Robotics Research 15:31 Combining Generative AI and Deep RL 21:24 Moravec's Paradox 25:33 Kitchen Robots 29:30 Simulation vs. Real-World Data 30:48 The Robot Olympics 36:31 The Physiological Reality of Embodiment 38:56 Controversies in the Robotics Community 44:18 What Makes a Great Researcher 48:27 How Businesses Should Prepare for Robotics 54:09 Tracking Progress Through Research Papers 57:02 The Next Step: Mid-Level Reasoning 1:02:00 The Kindest Thing

My conversation with Sergey Levine (Sergey Levine). Sergey is the co-founder of Physical Intelligence -- a company building foundation models that can control any robot to do any task in any environment. The company's thesis is that generality is more scalable than specialization, meaning that a model trained across many different robots and tasks will ultimately outperform any system built to do one thing well (eg, just wash dishes). Sergey is a researcher by background, but I think you will appreciate how practical and commercially grounded this conversation is. We discuss: - Why changing a diaper will be the last task a robot masters - The simulation v. real-world data debate - How multimodal LLMs give robots common sense - Moravec's Paradox + Robot Olympics - Why robots can do long-horizon tasks now - A realistic timeline for robots in our homes I should note that I am an investor in Physical Intelligence -- I made the investment because I believe it is one of the most important companies tackling the problem of robotics. Enjoy! Timestamps: 0:00 Intro 2:39 Defining Physical Intelligence 5:19 The Challenge of Building General Models 6:34 The Stakes and Future of General Purpose Robotics 8:15 Pros and Cons of Humanoid Robots 10:12 Historical Milestones in Robotics Research 15:31 Combining Generative AI and Deep RL 21:24 Moravec's Paradox 25:33 Kitchen Robots 29:30 Simulation vs. Real-World Data 30:48 The Robot Olympics 36:31 The Physiological Reality of Embodiment 38:56 Controversies in the Robotics Community 44:18 What Makes a Great Researcher 48:27 How Businesses Should Prepare for Robotics 54:09 Tracking Progress Through Research Papers 57:02 The Next Step: Mid-Level Reasoning 1:02:00 The Kindest Thing

Patrick OShaughnessy

133,833 просмотров • 4 месяцев назад