Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

A talk on verifiable environments for agents in biology 0:00 Where does experimental data come from? 1:41 Modern bio research organized around measurement 2:44 Data analysis is an executable substrate for science 3:23 Five years of learnings building products for pharma 4:13 Trying to use coding models for biology... show more

Kenny Workman

7,709 subscribers

89,760 views • 23 days ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

PantheonOS allows any biologist to perform complex data analyses of emerging single cell, multi-omics and spatial transcriptomics datasets end to end through AI agent and human collaboration. We are releasing six replays of use cases trajectories. Each "trajectory" is a complete end-to-end run, from prompt to analysis, to figures, and finally to report, and in fully inspectable and reproducible manner. See the six user cases in our PantheonOS Gallery: 1. 3D mouse embryo analysis: Tangram deconvolution and PyVista-based 3D visualization of E6 mouse embryo spatial transcriptomics data 2. 3D human fetal heart analysis: Spatial mapping of heart disease gene patterns using MERFISH data 3. Multi-omics spatial integration: Single cell multi-omics-to-spatial mapping with MOSCOT optimal transport 4. Gene panel design: 1000-plex immune-oncology MERFISH gene panel optimization 5. Cell segmentation benchmarking: Comparative evaluation of Cellpose-SAM, InstanSeg, StarDist, and other tools 6. Spatial disease biology: Ligand–receptor analysis of disease-associated tissue microenvironments We would love to hear how you can use PantheonOS for your research! Got an interesting agent run of your own and want to share? In the Pantheon UI, click Export Bundle (top-right of any chat) to package the full trajectory — chat history, code, figures, report — then submit it here:

PantheonOS allows any biologist to perform complex data analyses of emerging single cell, multi-omics and spatial transcriptomics datasets end to end through AI agent and human collaboration. We are releasing six replays of use cases trajectories. Each "trajectory" is a complete end-to-end run, from prompt to analysis, to figures, and finally to report, and in fully inspectable and reproducible manner. See the six user cases in our PantheonOS Gallery: 1. 3D mouse embryo analysis: Tangram deconvolution and PyVista-based 3D visualization of E6 mouse embryo spatial transcriptomics data 2. 3D human fetal heart analysis: Spatial mapping of heart disease gene patterns using MERFISH data 3. Multi-omics spatial integration: Single cell multi-omics-to-spatial mapping with MOSCOT optimal transport 4. Gene panel design: 1000-plex immune-oncology MERFISH gene panel optimization 5. Cell segmentation benchmarking: Comparative evaluation of Cellpose-SAM, InstanSeg, StarDist, and other tools 6. Spatial disease biology: Ligand–receptor analysis of disease-associated tissue microenvironments We would love to hear how you can use PantheonOS for your research! Got an interesting agent run of your own and want to share? In the Pantheon UI, click Export Bundle (top-right of any chat) to package the full trajectory — chat history, code, figures, report — then submit it here:

evo-devo

13,393 views • 3 months ago

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource

Bo Wang

199,747 views • 2 years ago

🌄 Beyond SWE: The Future of Long Horizon Environments A discussion with our founders about KellyBench, and the need for new environments that require agents to adapt over time and act under uncertainty. 0:00:17 What is KellyBench? 0:02:10 Openendedness, non-stationarity and continual learning 0:03:40 Analytical versus operational capabilities 0:04:13 Why are models bad at KellyBench? 0:05:39 Situational awareness in dynamic environments 0:06:37 Feature stability and real-world non-stationarity 0:07:07 The power of context 0:07:34 "The first principle is that you must not fool yourself" 0:08:20 Machiavelli, fortuna and the ability to adapt to change 0:09:23 How can models improve on evals like KellyBench? 0:10:12 Limitations of KellyBench: data availability and market odds timing 0:11:44 Implications beyond quant finance / sports betting 0:13:26 Civilisations as the ultimate time horizon 0:14:05 Would a mega prompt / better elicitation do much better on the benchmark? 0:14:52 What new types of capability is GR excited about? 0:17:48 Taste and the ability to pursue long-term goals even if they aren't immediately rewarding 0:18:56 Deep learning as an example of a method that took a long time to bear fruit 0:19:25 Optimism about the future of AI

🌄 Beyond SWE: The Future of Long Horizon Environments A discussion with our founders about KellyBench, and the need for new environments that require agents to adapt over time and act under uncertainty. 0:00:17 What is KellyBench? 0:02:10 Openendedness, non-stationarity and continual learning 0:03:40 Analytical versus operational capabilities 0:04:13 Why are models bad at KellyBench? 0:05:39 Situational awareness in dynamic environments 0:06:37 Feature stability and real-world non-stationarity 0:07:07 The power of context 0:07:34 "The first principle is that you must not fool yourself" 0:08:20 Machiavelli, fortuna and the ability to adapt to change 0:09:23 How can models improve on evals like KellyBench? 0:10:12 Limitations of KellyBench: data availability and market odds timing 0:11:44 Implications beyond quant finance / sports betting 0:13:26 Civilisations as the ultimate time horizon 0:14:05 Would a mega prompt / better elicitation do much better on the benchmark? 0:14:52 What new types of capability is GR excited about? 0:17:48 Taste and the ability to pursue long-term goals even if they aren't immediately rewarding 0:18:56 Deep learning as an example of a method that took a long time to bear fruit 0:19:25 Optimism about the future of AI

General Reasoning

15,585 views • 4 months ago

When Mudith Jayasekara and I met Gabe Pereyra, we were expecting just another vanilla intro call and instead had the best yarn about research, the state of LLMs, and where intelligence is actually heading. It's rare to meet a founder this deep in the weeds who's also building for one of the most important verticals in this new age of intelligence So it was awesome to sit down with Gabe for an extended discussion on what it take to build agents that can reliably complete work over hours, days, or even longer? We talked about why agents today struggle with search and long context windows and how techniques like KV-cache compaction, synthetic data, and continual learning could help. 0:00 Introduction 0:36 Getting legal agents to review the whole data room 2:08 Data rooms larger than any context window 5:28 How far open-source models can go 7:58 Where specialist models fit in legal AI 10:59 Training legal models when client data is off-limits 13:06 Teaching a model how a law firm works 13:59 What belongs in context vs. model weights 15:36 From firm-wide AI to a model for every lawyer 18:37 What training adds beyond retrieving the right cases 20:26 Why context windows have plateaued 24:01 How models could learn continuously on the job 26:12 Can AI recursively improve AI research? 27:07 Research agents can run experiments but not choose them 30:00 Why open-ended research is hard to train 33:47 Why deployment, not intelligence, is the bottleneck 35:08 The cost of frontier intelligence 36:59 Different neolabs, different paths to intelligence 39:26 Using open datasets to compare research methods 41:13 Conclusion

When Mudith Jayasekara and I met Gabe Pereyra, we were expecting just another vanilla intro call and instead had the best yarn about research, the state of LLMs, and where intelligence is actually heading. It's rare to meet a founder this deep in the weeds who's also building for one of the most important verticals in this new age of intelligence So it was awesome to sit down with Gabe for an extended discussion on what it take to build agents that can reliably complete work over hours, days, or even longer? We talked about why agents today struggle with search and long context windows and how techniques like KV-cache compaction, synthetic data, and continual learning could help. 0:00 Introduction 0:36 Getting legal agents to review the whole data room 2:08 Data rooms larger than any context window 5:28 How far open-source models can go 7:58 Where specialist models fit in legal AI 10:59 Training legal models when client data is off-limits 13:06 Teaching a model how a law firm works 13:59 What belongs in context vs. model weights 15:36 From firm-wide AI to a model for every lawyer 18:37 What training adds beyond retrieving the right cases 20:26 Why context windows have plateaued 24:01 How models could learn continuously on the job 26:12 Can AI recursively improve AI research? 27:07 Research agents can run experiments but not choose them 30:00 Why open-ended research is hard to train 33:47 Why deployment, not intelligence, is the bottleneck 35:08 The cost of frontier intelligence 36:59 Different neolabs, different paths to intelligence 39:26 Using open datasets to compare research methods 41:13 Conclusion

Charlie O'Neill

91,088 views • 1 month ago

After many months, a lot of model releases and even more Amp releases, here it is: A new episode of Raising An Agent with Quinn Slack and Thorsten Ball. They discuss: - Why Gemini 3 might have been the inflection point, not Opus 4.5 - What changed over Christmas and why everybody talks about agents now - "The assistant is dead, long live the factory" - How codebases need to adapt for agents - Why small/local models don't give you an impression of the frontier - What does it mean for a codebase to be agent-friendly? - New things coming to Amp Enjoy! And happy hacking! Timestamps: 1:51 The Impact of Gemini 3 and Opus 4.5 5:28 Shifting Perceptions of AI Agents 10:17 Adapting Codebases for AI Efficiency 14:48 The Future of Software Development with AI Agents 19:29 Challenges and Opportunities Ahead 24:00 The Importance of Frontier Models 25:12 Adapting to AI-Driven Workflows 26:57 The Future of Coding Agents 28:07 Building an Agentic Code Base 30:22 Rethinking Development Tools 31:44 Amp's Evolution and Team Dynamics 33:00 Discord Q&A 39:18 Teasing New Features: Deep

After many months, a lot of model releases and even more Amp releases, here it is: A new episode of Raising An Agent with Quinn Slack and Thorsten Ball. They discuss: - Why Gemini 3 might have been the inflection point, not Opus 4.5 - What changed over Christmas and why everybody talks about agents now - "The assistant is dead, long live the factory" - How codebases need to adapt for agents - Why small/local models don't give you an impression of the frontier - What does it mean for a codebase to be agent-friendly? - New things coming to Amp Enjoy! And happy hacking! Timestamps: 1:51 The Impact of Gemini 3 and Opus 4.5 5:28 Shifting Perceptions of AI Agents 10:17 Adapting Codebases for AI Efficiency 14:48 The Future of Software Development with AI Agents 19:29 Challenges and Opportunities Ahead 24:00 The Importance of Frontier Models 25:12 Adapting to AI-Driven Workflows 26:57 The Future of Coding Agents 28:07 Building an Agentic Code Base 30:22 Rethinking Development Tools 31:44 Amp's Evolution and Team Dynamics 33:00 Discord Q&A 39:18 Teasing New Features: Deep

Amp

36,150 views • 7 months ago

Today, we're joined by Aakanksha Chowdhery, member of technical staff at Reflection, to explore the fundamental shifts required to build true agentic AI. While the industry has largely focused on post-training techniques to improve reasoning, Aakanksha draws on her experience leading pre-training efforts for Google’s PaLM and early Gemini models to argue that pre-training itself must be rethought to move beyond static benchmarks. We explore the limitations of next-token prediction for multi-step workflows and examine how attention mechanisms, loss objectives, and training data must evolve to support long-form reasoning and planning. Aakanksha shares insights on the difference between context retrieval and actual reasoning, the importance of "trajectory" training data, and why scaling remains essential for discovering emergent agentic capabilities like error recovery and dynamic tool learning. 🗒️ For the full list of resources for this episode, visit the show notes page: 📖 CHAPTERS =============================== 00:00 - Introduction 02:26 - Reflection 04:54 - Limitations of post-training for building agents 07:31 - Rethinking pre-training in agents 10:51 - Scaling 11:27 - Evolving attention mechanisms for agentic capabilities 12:39 - Memory as a tool 14:13 - Loss objectives and training data 15:50 - Fine-tuning loss in agent performance 19:37 - Training data 21:29 - Augmenting dominant training data source 24:11 - Overcoming challenges in training on synthetic data 25:47 - Benchmarks 30:44 - Scaling laws in large models versus small models 33:20 - Long-form versus short-form reasoning 37:57 - Agent’s ability to recover from failure 40:15 - Hallucinations and failure recovery 43:53 - Tool use in agents 46:38 - Coding agents 48:37 - How researchers can contribute to agentic AI

Today, we're joined by Aakanksha Chowdhery, member of technical staff at Reflection, to explore the fundamental shifts required to build true agentic AI. While the industry has largely focused on post-training techniques to improve reasoning, Aakanksha draws on her experience leading pre-training efforts for Google’s PaLM and early Gemini models to argue that pre-training itself must be rethought to move beyond static benchmarks. We explore the limitations of next-token prediction for multi-step workflows and examine how attention mechanisms, loss objectives, and training data must evolve to support long-form reasoning and planning. Aakanksha shares insights on the difference between context retrieval and actual reasoning, the importance of "trajectory" training data, and why scaling remains essential for discovering emergent agentic capabilities like error recovery and dynamic tool learning. 🗒️ For the full list of resources for this episode, visit the show notes page: 📖 CHAPTERS =============================== 00:00 - Introduction 02:26 - Reflection 04:54 - Limitations of post-training for building agents 07:31 - Rethinking pre-training in agents 10:51 - Scaling 11:27 - Evolving attention mechanisms for agentic capabilities 12:39 - Memory as a tool 14:13 - Loss objectives and training data 15:50 - Fine-tuning loss in agent performance 19:37 - Training data 21:29 - Augmenting dominant training data source 24:11 - Overcoming challenges in training on synthetic data 25:47 - Benchmarks 30:44 - Scaling laws in large models versus small models 33:20 - Long-form versus short-form reasoning 37:57 - Agent’s ability to recover from failure 40:15 - Hallucinations and failure recovery 43:53 - Tool use in agents 46:38 - Coding agents 48:37 - How researchers can contribute to agentic AI

The TWIML AI Podcast

44,888 views • 8 months ago

.george church is the godfather of modern synthetic biology. He thinks that the last few decades of progress have put us on the verge of the big payoffs: deaging, deextinction, biobots that combine the best of human & natural engineering, and (unfortunately) weaponized mirror life. 0:00:00 – Aging solved by 2050 0:07:37 – Finding the master switch for any trait 0:19:50 – Weaponized mirror life 0:30:40 – Why hasn’t sequencing/synthesis led to biotech revolution? 0:50:26 – Impact of AGI on biology research progress 1:00:35 – Biobots that use the best of biological and human engineering 1:05:09 – Odds of life in universe 1:09:57 – Is DNA the ultimate data storage? 1:13:55 – Curing rare diseases with genetic counseling 1:22:23 – NIH & NSF budget cuts 1:25:26 – How one lab spawned 100 biotech companies Watch full episode on YouTube, Spotify, Apple Podcasts, etc. Transcript at my website dwarkesh dot com. Enjoy!

.george church is the godfather of modern synthetic biology. He thinks that the last few decades of progress have put us on the verge of the big payoffs: deaging, deextinction, biobots that combine the best of human & natural engineering, and (unfortunately) weaponized mirror life. 0:00:00 – Aging solved by 2050 0:07:37 – Finding the master switch for any trait 0:19:50 – Weaponized mirror life 0:30:40 – Why hasn’t sequencing/synthesis led to biotech revolution? 0:50:26 – Impact of AGI on biology research progress 1:00:35 – Biobots that use the best of biological and human engineering 1:05:09 – Odds of life in universe 1:09:57 – Is DNA the ultimate data storage? 1:13:55 – Curing rare diseases with genetic counseling 1:22:23 – NIH & NSF budget cuts 1:25:26 – How one lab spawned 100 biotech companies Watch full episode on YouTube, Spotify, Apple Podcasts, etc. Transcript at my website dwarkesh dot com. Enjoy!

Dwarkesh Patel

85,882 views • 1 year ago

Reconstructing the world in 3D from random internet photos is a deep rabbit hole. In 2008, Photosynth proved that 3D reconstruction of smaller monuments was possible. The next year, "Building Rome in a Day" proved it could be done for entire cities. Since then, there’s been a wild lineage of research from early point clouds to modern NeRFs and 3DGS --with a similar cast of characters appearing along the way. In this deep dive, we'll break it all down, and explore why IARPA is funding what comes next. 0:00 The Internet's 3D Model of the World 0:40 Building Rome in a Day (2009) 2:09 The Long Tail Problem 3:18 MegaDepth and the Mannequin Challenge 4:44 NeRFs and 3D Gaussian Splatting 7:19 MegaScenes, Doppelgangers, and Foundation Models 10:00 MegaDepth-X Punches Through the Wall 11:26 Why IARPA Is Funding What's Next 13:51 The 4D Frontier and the Sensorium

Reconstructing the world in 3D from random internet photos is a deep rabbit hole. In 2008, Photosynth proved that 3D reconstruction of smaller monuments was possible. The next year, "Building Rome in a Day" proved it could be done for entire cities. Since then, there’s been a wild lineage of research from early point clouds to modern NeRFs and 3DGS --with a similar cast of characters appearing along the way. In this deep dive, we'll break it all down, and explore why IARPA is funding what comes next. 0:00 The Internet's 3D Model of the World 0:40 Building Rome in a Day (2009) 2:09 The Long Tail Problem 3:18 MegaDepth and the Mannequin Challenge 4:44 NeRFs and 3D Gaussian Splatting 7:19 MegaScenes, Doppelgangers, and Foundation Models 10:00 MegaDepth-X Punches Through the Wall 11:26 Why IARPA Is Funding What's Next 13:51 The 4D Frontier and the Sensorium

Bilawal Sidhu

51,292 views • 12 days ago

New Infinite Loops with Saloni Dattani (Saloni)! There may be medical breakthroughs in the pipeline right now that work — but won’t reach patients for another decade. Saloni Dattani joins us to explain the hidden bottleneck slowing the future of medicine and why biology may be one of the most important fields to work in today. Listen now. TIMESTAMPS: 0:00 Intro 2:07 The Problem with Health Data 14:44 Clinical Trials Are the Bottleneck 25:05 AI and Drug Discovery 35:00 Why Science Journalism Fails 45:41 Medicine’s Next Great Leap 50:20 Fixing Medical Market Failures 57:10 The Leukemia Trial Breakthrough 1:06:18 Why Failed Research Matters 1:12:51 Science Is Never Settled

New Infinite Loops with Saloni Dattani (Saloni)! There may be medical breakthroughs in the pipeline right now that work — but won’t reach patients for another decade. Saloni Dattani joins us to explain the hidden bottleneck slowing the future of medicine and why biology may be one of the most important fields to work in today. Listen now. TIMESTAMPS: 0:00 Intro 2:07 The Problem with Health Data 14:44 Clinical Trials Are the Bottleneck 25:05 AI and Drug Discovery 35:00 Why Science Journalism Fails 45:41 Medicine’s Next Great Leap 50:20 Fixing Medical Market Failures 57:10 The Leukemia Trial Breakthrough 1:06:18 Why Failed Research Matters 1:12:51 Science Is Never Settled

Infinite Loops 🎙

11,818 views • 3 months ago

Get the inside story on the development of Gemini's coding capabilities. Listen as the product and research leads for Gemini share their philosophy on what makes a great coding model, the impact of "vibe coding," and the future of programming languages with Logan Kilpatrick, Connie Fan and Danny Tarlow. Timecodes: 0:00 Intro 1:10 Defining Early Coding Goals 6:23 Ingredients of a Great Coding Model 9:28 Adapting to Developer Workflows 11:40 The Rise of Vibe Coding 14:43 Code as a Reasoning Tool 17:20 Code as a Universal Solver 20:47 Evaluating Coding Models 24:30 Leveraging Internal Googler Feedback 26:52 Winning Over AI Skeptics 28:04 Performance Across Programming Languages 33:05 The Future of Programming Languages 36:16 Strategies for Large Codebases 41:06 Hill Climbing New Benchmarks 42:46 Short-Term Improvements 44:42 Model Style and Taste 47:43 2.5 Pro’s Breakthrough 51:06 Early AI Coding Experiences 56:19 Specialist vs. Generalist Models

Get the inside story on the development of Gemini's coding capabilities. Listen as the product and research leads for Gemini share their philosophy on what makes a great coding model, the impact of "vibe coding," and the future of programming languages with Logan Kilpatrick, Connie Fan and Danny Tarlow. Timecodes: 0:00 Intro 1:10 Defining Early Coding Goals 6:23 Ingredients of a Great Coding Model 9:28 Adapting to Developer Workflows 11:40 The Rise of Vibe Coding 14:43 Code as a Reasoning Tool 17:20 Code as a Universal Solver 20:47 Evaluating Coding Models 24:30 Leveraging Internal Googler Feedback 26:52 Winning Over AI Skeptics 28:04 Performance Across Programming Languages 33:05 The Future of Programming Languages 36:16 Strategies for Large Codebases 41:06 Hill Climbing New Benchmarks 42:46 Short-Term Improvements 44:42 Model Style and Taste 47:43 2.5 Pro’s Breakthrough 51:06 Early AI Coding Experiences 56:19 Specialist vs. Generalist Models

Google AI Developers

65,474 views • 1 year ago

A model is only as good as its data, and we’ve long since exhausted the internet. From here on out, model progress is gated by data production. ’s Brendan (can/do) joined us at our Sovereign AI event to talk about how RL environments get built, and why your data might be your real moat: 00:00 Introduction 00:47 A short history of the data market: crowdsourcing to agentic data 02:29 What an RL environment is: worlds, apps, tasks 03:57 Why only humans can measure the frontier 05:35 Building verifiers is the hard part 06:44 Walkthrough: a real legal RL environment 08:18 Leaderboards — and what open weights change 09:45 Post-training results on Apex Agents 11:17 Three ways companies buy data 12:49 Q&A: How do you price data? 14:17 Q&A: What "data quality" actually means 16:42 Q&A: The misunderstanding about synthetic data 18:17 Q&A: Why RL environments now — and what comes after 21:20 Q&A: Can you scale rubric generation with models? 23:00 Q&A: RL environments for cyber defense 25:33 Q&A: Build data in-house or partner?

A model is only as good as its data, and we’ve long since exhausted the internet. From here on out, model progress is gated by data production. ’s Brendan (can/do) joined us at our Sovereign AI event to talk about how RL environments get built, and why your data might be your real moat: 00:00 Introduction 00:47 A short history of the data market: crowdsourcing to agentic data 02:29 What an RL environment is: worlds, apps, tasks 03:57 Why only humans can measure the frontier 05:35 Building verifiers is the hard part 06:44 Walkthrough: a real legal RL environment 08:18 Leaderboards — and what open weights change 09:45 Post-training results on Apex Agents 11:17 Three ways companies buy data 12:49 Q&A: How do you price data? 14:17 Q&A: What "data quality" actually means 16:42 Q&A: The misunderstanding about synthetic data 18:17 Q&A: Why RL environments now — and what comes after 21:20 Q&A: Can you scale rubric generation with models? 23:00 Q&A: RL environments for cyber defense 25:33 Q&A: Build data in-house or partner?

Pat Grady

65,287 views • 12 days ago

BREAKING: AI has eaten the Internet. Data labeling is so over. & $30 trillion of human work is on the verge of automation. Inside The $2.2B AI Research Accelerator, Turing Founder & CEO, Jonathan Siddharth (), joins Sourcery to break down the severe power shift in AI training: from commodity data labeling → expert research Positioning Turing apart from AI data providers like Scale AI, Mercor, & Surge. (00:00) AI Ate The Internet (00:49) Training Superintelligence: the race to AGI (02:31) Viral tweet (03:24) What Turing actually does (04:43) The internet data is “used up” — where will new data come from? (05:34) Four pillars of superintelligence: multimodality, reasoning, tool use, coding (06:07) Automating $30T of global knowledge work (09:18) The $1B revenue opportunity (10:59) Why Turing is a research-first accelerator, not a data labeler (13:45) Jonathan’s Stanford AI Lab roots & founding DNA (17:57) How models are built: pre-training vs. post-training (20:14) RLHF, reinforcement learning, & “breaking the models” (25:19) GPT-5 and the myth of rapid takeoff (30:46) Safety debates and human-in-the-loop systems (34:53) Closing Enterprise Gap: finance, insurance, & pharma (39:23) Why proprietary enterprise data is the next moat in AI

BREAKING: AI has eaten the Internet. Data labeling is so over. & $30 trillion of human work is on the verge of automation. Inside The $2.2B AI Research Accelerator, Turing Founder & CEO, Jonathan Siddharth (), joins Sourcery to break down the severe power shift in AI training: from commodity data labeling → expert research Positioning Turing apart from AI data providers like Scale AI, Mercor, & Surge. (00:00) AI Ate The Internet (00:49) Training Superintelligence: the race to AGI (02:31) Viral tweet (03:24) What Turing actually does (04:43) The internet data is “used up” — where will new data come from? (05:34) Four pillars of superintelligence: multimodality, reasoning, tool use, coding (06:07) Automating $30T of global knowledge work (09:18) The $1B revenue opportunity (10:59) Why Turing is a research-first accelerator, not a data labeler (13:45) Jonathan’s Stanford AI Lab roots & founding DNA (17:57) How models are built: pre-training vs. post-training (20:14) RLHF, reinforcement learning, & “breaking the models” (25:19) GPT-5 and the myth of rapid takeoff (30:46) Safety debates and human-in-the-loop systems (34:53) Closing Enterprise Gap: finance, insurance, & pharma (39:23) Why proprietary enterprise data is the next moat in AI

Molly O’Shea

70,007 views • 10 months ago

Here's my conversation with Ed Barnhart, an archaeologist specializing in ancient civilizations of the Americas. We talk about the Mayan Civilization, Aztec Empire, Inca Empire, and the lost civilizations of South America and the Amazon jungle. South America is one of the cradles of human civilization. Studying this ancient history lays bare the power, beauty, and dangers of human nature manifested in many of its forms across thousands of years. It's here on X in full, and is up everywhere else. Links in comment. Timestamps: 0:00 - Introduction 1:39 - Lost civilizations 8:43 - Hunter-gatherers 12:16 - First humans in the Americas 22:07 - South America 27:36 - Pyramids 34:40 - Religion 47:44 - Shamanism 49:41 - Ayahuasca 55:54 - Lost City of Z 1:00:48 - Graham Hancock 1:07:51 - Uncontacted tribes 1:13:51 - Maya civilization 1:29:40 - Mayan calendar 1:44:57 - Flood myths 2:13:25 - Aztecs 2:30:52 - Inca Empire 2:48:52 - Early humans in North America 2:54:50 - Columbus 2:59:26 - Vikings 3:03:35 - Aliens 3:08:02 - Earth in 10,000 years 3:24:12 - Hope for the future

Here's my conversation with Ed Barnhart, an archaeologist specializing in ancient civilizations of the Americas. We talk about the Mayan Civilization, Aztec Empire, Inca Empire, and the lost civilizations of South America and the Amazon jungle. South America is one of the cradles of human civilization. Studying this ancient history lays bare the power, beauty, and dangers of human nature manifested in many of its forms across thousands of years. It's here on X in full, and is up everywhere else. Links in comment. Timestamps: 0:00 - Introduction 1:39 - Lost civilizations 8:43 - Hunter-gatherers 12:16 - First humans in the Americas 22:07 - South America 27:36 - Pyramids 34:40 - Religion 47:44 - Shamanism 49:41 - Ayahuasca 55:54 - Lost City of Z 1:00:48 - Graham Hancock 1:07:51 - Uncontacted tribes 1:13:51 - Maya civilization 1:29:40 - Mayan calendar 1:44:57 - Flood myths 2:13:25 - Aztecs 2:30:52 - Inca Empire 2:48:52 - Early humans in North America 2:54:50 - Columbus 2:59:26 - Vikings 3:03:35 - Aliens 3:08:02 - Earth in 10,000 years 3:24:12 - Hope for the future

Lex Fridman

309,398 views • 1 year ago

The Bitter Lesson for Biology — Adam Green on Scaling Laws for Virtual Cells In this episode, the founder of Markov Biosciences, Adam Green, explains the "bitter lesson" for biology, the idea borrowed from Richard Sutton that large unbiased datasets and the right training objective tend to outcompete models with hard-coded rules and human priors. We talk a lot about virtual cells, sources of bias in data, and his evidence that virtual cells pre-trained on plain observational data show clean scaling laws, getting monotonically better at predicting unseen perturbations as the models grow, and beating a state-of-the-art model built specifically for that task. Our interview was about 3 hours long, but cut to 1.5 hours. My main goal was to understand Adam's perspective on virtual cells and his goal to "solve biology." Search for "The New Biology" on YouTube, Spotify, and Apple Podcasts. Timestamps: 00:00 — Cold open 01:58 — Clinical predictions from a virtual cell 05:38 — What is a "virtual cell"? 08:01 — The problems with single-cell RNA-seq 11:31 — The urns analogy 19:54 — Observational vs. perturbation 23:29 — The bitter lesson for biology 29:06 — Geometric Plackett-Luce 38:27 — Ablations and loss function 47:23 — Cells as specimens 59:26 — Antibody-Drug conjugates 1:11:16 — Will we ever understand the cell fully?

The Bitter Lesson for Biology — Adam Green on Scaling Laws for Virtual Cells In this episode, the founder of Markov Biosciences, Adam Green, explains the "bitter lesson" for biology, the idea borrowed from Richard Sutton that large unbiased datasets and the right training objective tend to outcompete models with hard-coded rules and human priors. We talk a lot about virtual cells, sources of bias in data, and his evidence that virtual cells pre-trained on plain observational data show clean scaling laws, getting monotonically better at predicting unseen perturbations as the models grow, and beating a state-of-the-art model built specifically for that task. Our interview was about 3 hours long, but cut to 1.5 hours. My main goal was to understand Adam's perspective on virtual cells and his goal to "solve biology." Search for "The New Biology" on YouTube, Spotify, and Apple Podcasts. Timestamps: 00:00 — Cold open 01:58 — Clinical predictions from a virtual cell 05:38 — What is a "virtual cell"? 08:01 — The problems with single-cell RNA-seq 11:31 — The urns analogy 19:54 — Observational vs. perturbation 23:29 — The bitter lesson for biology 29:06 — Geometric Plackett-Luce 38:27 — Ablations and loss function 47:23 — Cells as specimens 59:26 — Antibody-Drug conjugates 1:11:16 — Will we ever understand the cell fully?

Niko McCarty.

27,869 views • 2 months ago

The tensorqt in SF episode is here. This is special. Just being terminally online on this app - post ideas, meet mutuals, raise funds, and build a frontier company. This has been Francesco's journey building infra for autonomous research at Paradigma. 0:00:00 - INTRO 0:01:03 - Dropping out of PhD, Saturation in Academia, Building from EU 0:13:15 - Autoresearch, Idea behind Flywheel, ex-OpenAI Co-founder 0:19:18 - Evals for Autoreserch and Research Taste 0:25:00 - How a Research Lab uses Flywheel? 0:31:15 - Thesis and Autoresearch Adoption in Top 5 Labs 0:41:00 - Working around with Flywheel, Combinatorial Explosion 1:02:03 - Science as Public Graph, Deriving New Research Loop, Quick Fire 1:06:23 - Toughest Autoresearch Challenge as of Today, Future Directions 1:11:53 - Automating Science: Sakana and Deepmind, Trust Problem? 1:21:07 - Business Model, Automating R&D, Model Capabilities 1:27:20 - Dropout of college to live on frontier!

The tensorqt in SF episode is here. This is special. Just being terminally online on this app - post ideas, meet mutuals, raise funds, and build a frontier company. This has been Francesco's journey building infra for autonomous research at Paradigma. 0:00:00 - INTRO 0:01:03 - Dropping out of PhD, Saturation in Academia, Building from EU 0:13:15 - Autoresearch, Idea behind Flywheel, ex-OpenAI Co-founder 0:19:18 - Evals for Autoreserch and Research Taste 0:25:00 - How a Research Lab uses Flywheel? 0:31:15 - Thesis and Autoresearch Adoption in Top 5 Labs 0:41:00 - Working around with Flywheel, Combinatorial Explosion 1:02:03 - Science as Public Graph, Deriving New Research Loop, Quick Fire 1:06:23 - Toughest Autoresearch Challenge as of Today, Future Directions 1:11:53 - Automating Science: Sakana and Deepmind, Trust Problem? 1:21:07 - Business Model, Automating R&D, Model Capabilities 1:27:20 - Dropout of college to live on frontier!

himanshu

14,913 views • 4 months ago

🚀 Introducing PantheonOS ( A Fully Open-Source Agent OS for Science PantheonOS began as a research project in my Stanford lab and has since evolved into a vision to redefine data science in the era of AI—starting with computational biology, especially single-cell and spatial genomics. PantheonOS is a general agent platform built from the ground up. It is arguably the first distributed agent framework designed for scientific data analysis. 🔑 Key Features 1. Multi-Agent Collaboration – Built-in paradigms for distributed, cross-machine cooperation among agents and toolsets. 2. Native Toolset Support – Python, R, Julia, LaTeX, and more—designed for real scientific workflows. 3. Modular & Extensible – Developer-friendly design with shallow wrappers, plus LLM-driven toolset generation. 4. Evolvable Agents – Capable of evolving large-scale code projects to achieve superhuman performance (e.g., evolving upon the original Harmony [I Korsunsky, 2019, Nature Biotechnology] and Scanorama [BL Hie, 2019, Nature Biotechnology] implementations), and even evolving the system itself to adapt to new fields. 🎉 Stepwise Release Strategy We’re releasing PantheonOS in stages: Pantheon-CLI (today!), followed by Pantheon-Lab, Pantheon-Notebook, Pantheon-Slack, and more. 🌟 Pantheon-CLI Highlights - We're not just building another CLI tool. We're defining how scientists will interact with data in the AI era. - Open, Powerful, Python-First – The first fully open-source, endlessly extendable scientific “vibe analysis” framework. - Mixed Programming Magic – Combine Python, natural language, R, or Julia—seamlessly in the same environment. - PhD-Level Assistant – A command-line agent for complex real-world genomics and beyond, handling workflows at the PhD level. - Privacy by Design – Run entirely offline with local LLMs—your data never leaves your computer. ✅ Proven Applications (10 Demonstrations) Computational biology: 1. ATAC-seq: From raw reads to peak matrix 2. RNA-seq: From raw reads to expression matrix 3. Complex single-cell workflows (PhD-level) 4. Hybrid natural language + R for Seurat annotation 5. Learning from web tutorials + invoking single-cell foundation models 6. Cell segmentation on 10x Genomics HD Visium data And beyond: 7. Mixed Python & R programming examples 8. Molecular docking & structural analysis 9. Exploratory factor analysis for behavioral survey data 10. Customer segmentation & finance analytics 🌐 Learn More & Get Started Website: Pantheon-CLI Documentation: GitHub Repo: 💬 Join our community: PantheonOS Slack: PantheonOS Discord:

🚀 Introducing PantheonOS ( A Fully Open-Source Agent OS for Science PantheonOS began as a research project in my Stanford lab and has since evolved into a vision to redefine data science in the era of AI—starting with computational biology, especially single-cell and spatial genomics. PantheonOS is a general agent platform built from the ground up. It is arguably the first distributed agent framework designed for scientific data analysis. 🔑 Key Features 1. Multi-Agent Collaboration – Built-in paradigms for distributed, cross-machine cooperation among agents and toolsets. 2. Native Toolset Support – Python, R, Julia, LaTeX, and more—designed for real scientific workflows. 3. Modular & Extensible – Developer-friendly design with shallow wrappers, plus LLM-driven toolset generation. 4. Evolvable Agents – Capable of evolving large-scale code projects to achieve superhuman performance (e.g., evolving upon the original Harmony [I Korsunsky, 2019, Nature Biotechnology] and Scanorama [BL Hie, 2019, Nature Biotechnology] implementations), and even evolving the system itself to adapt to new fields. 🎉 Stepwise Release Strategy We’re releasing PantheonOS in stages: Pantheon-CLI (today!), followed by Pantheon-Lab, Pantheon-Notebook, Pantheon-Slack, and more. 🌟 Pantheon-CLI Highlights - We're not just building another CLI tool. We're defining how scientists will interact with data in the AI era. - Open, Powerful, Python-First – The first fully open-source, endlessly extendable scientific “vibe analysis” framework. - Mixed Programming Magic – Combine Python, natural language, R, or Julia—seamlessly in the same environment. - PhD-Level Assistant – A command-line agent for complex real-world genomics and beyond, handling workflows at the PhD level. - Privacy by Design – Run entirely offline with local LLMs—your data never leaves your computer. ✅ Proven Applications (10 Demonstrations) Computational biology: 1. ATAC-seq: From raw reads to peak matrix 2. RNA-seq: From raw reads to expression matrix 3. Complex single-cell workflows (PhD-level) 4. Hybrid natural language + R for Seurat annotation 5. Learning from web tutorials + invoking single-cell foundation models 6. Cell segmentation on 10x Genomics HD Visium data And beyond: 7. Mixed Python & R programming examples 8. Molecular docking & structural analysis 9. Exploratory factor analysis for behavioral survey data 10. Customer segmentation & finance analytics 🌐 Learn More & Get Started Website: Pantheon-CLI Documentation: GitHub Repo: 💬 Join our community: PantheonOS Slack: PantheonOS Discord:

evo-devo

17,410 views • 1 year ago

My conversation with Jesse Zhang Decagon is building a customer service agent with AI, with plans for much more than that. It’s one of the more successful early AI applications, with tons of lessons for the rest of the market. I love how Jesse developed which problem to solve first, a set of lessons applicable to all entrepreneurs Enjoy Timestamps: 0:00 Intro 2:00 Competing in the AI Era 11:35 Finding Product-Market Fit 14:25 Anatomy of a Great Customer Interview 17:37 Everyone’s Wrong Until You Talk to Customers 19:25 AI Agents Replacing Human Labor 21:55 Why Customer Service 23:30 Coding vs. Customer Service 35:44 The Future of Voice and the Human UI 38:54 Why Voice-to-Voice Is So Hard 43:18 Learning from Customer Data 44:36 How CEOs Should Think About AI 47:06 The ROI of Coding and Productivity Tools 48:20 Every Brand Will Talk to You Through AI 56:00 Recruiting, Capital, and Demand 1:00:42 Investor Frenzy 1:22:17 The Long-Term Vision for Decagon 1:25:05 Kindest Thing

Patrick OShaughnessy

49,418 views • 10 months ago

I spoke with Nick Grossman about his "Rebel Alliance" thesis for AI. He believes the industry is too big for one or two companies to dominate. We also discuss USV's internal multi-agent platform, model routing economics, AI's growing role in venture capital and financial markets, data privacy risks, and what happens to jobs and society as agents get smarter. YouTube: Spotify: Apple: TIMESTAMPS: 0:00 - Intro 0:42 - The "Rebel Alliance" thesis: why one company won't dominate AI 7:15 - General-purpose vs. specialized models in USV's portfolio 8:40 - Compute costs & the rise of model routing 11:14 - Inside USV's internal multi-agent system 16:25 - AI's role in venture capital & autonomous investing 22:46 - How AI reshapes market signals & information edges 27:02 - Apps building models & the fight over your data 35:26 - Five-year outlook for the AI stack 37:53 - US vs. China: the "philosophy" of AI models 40:26 - Personalized AI & where the Rebel Alliance thesis plays 43:49 - The dark side: jobs, data centers & society 47:40 - AI as a personal superpower & neural implants

I spoke with Nick Grossman about his "Rebel Alliance" thesis for AI. He believes the industry is too big for one or two companies to dominate. We also discuss USV's internal multi-agent platform, model routing economics, AI's growing role in venture capital and financial markets, data privacy risks, and what happens to jobs and society as agents get smarter. YouTube: Spotify: Apple: TIMESTAMPS: 0:00 - Intro 0:42 - The "Rebel Alliance" thesis: why one company won't dominate AI 7:15 - General-purpose vs. specialized models in USV's portfolio 8:40 - Compute costs & the rise of model routing 11:14 - Inside USV's internal multi-agent system 16:25 - AI's role in venture capital & autonomous investing 22:46 - How AI reshapes market signals & information edges 27:02 - Apps building models & the fight over your data 35:26 - Five-year outlook for the AI stack 37:53 - US vs. China: the "philosophy" of AI models 40:26 - Personalized AI & where the Rebel Alliance thesis plays 43:49 - The dark side: jobs, data centers & society 47:40 - AI as a personal superpower & neural implants

Anthony Pompliano 🌪

60,709 views • 1 month ago