Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

OpenAI Lead Researcher, Lukasz Kaiser: Reasoning models already handle most computer tasks — writing, coding, clicking, though still quirky Competition among labs is speeding progress, with research improvements in the pipeline "before 2030, much desk work will shift to AI"

slow_developer's profile picture

Haider.

57,339 subscribers

82,622 Aufrufe • vor 9 Monaten •via X (Twitter)

Wissenschaft & TechnologieNachrichten & Politik
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

New AI Podcast 🔥 | Lukasz Kaiser OpenAI Today is a very special day for us. We have Lukasz Kaiser, Lead Researcher at OpenAI, on our AI podcast, one of the co-creators of the Transformers and the groundbreaking paper “Attention Is All You Need” that started it all. We could easily say he is one of the creators of modern artificial intelligence. It was a true pleasure to talk with one of the brightest minds in the world of AI. Link pinned in the comments below. Thanks Lukasz Kaiser
2:23

New AI Podcast 🔥 | Lukasz Kaiser OpenAI Today is a very special day for us. We have Lukasz Kaiser, Lead Researcher at OpenAI, on our AI podcast, one of the co-creators of the Transformers and the groundbreaking paper “Attention Is All You Need” that started it all. We could easily say he is one of the creators of modern artificial intelligence. It was a true pleasure to talk with one of the brightest minds in the world of AI. Link pinned in the comments below. Thanks Lukasz Kaiser

Jon Hernandez

98,966 Aufrufe • vor 9 Monaten

Sam Altman says the next big thrust is making models handle much longer tasks We have already moved from models handling 5-second coding tasks with GPT-3.5 to 5-hour tasks with GPT-5 The goal now is to integrate enterprise context so AI can handle tasks that require months or years
1:33

Sam Altman says the next big thrust is making models handle much longer tasks We have already moved from models handling 5-second coding tasks with GPT-3.5 to 5-hour tasks with GPT-5 The goal now is to integrate enterprise context so AI can handle tasks that require months or years

Haider.

122,717 Aufrufe • vor 8 Monaten

I sat down with Lukasz Kaiser to get into whether the architecture he helped invent is actually enough, and what's next in generalization, coding agents, RL and more. Lukasz co-authored "Attention Is All You Need," the paper that introduced the transformer and worked on reasoning models at OpenAI so he’s been a key part of major shifts in the field. We hit on: ▪️ The case for and against a new architecture coming after the transformer ▪️ What’s required for model generalization in the physical world ▪️ How much coding agents have improved his AI research productivity ▪️ The next domains for RL ▪️ Why Anthropic initially won coding ▪️ Future research directions he’s excited about 0:00 Intro 1:12 Transformers vs. Human Learning 8:37 How Do We Get Physical World Generalization? 10:52 What Comes After Transformers 13:59 How Much Have Agents Improved Lukasz's AI Research Productivity? 17:21 How Close Is an AI Research Intern? 26:06 RL Beyond Verifiable Tasks 35:38 App Companies: Build Models or Lean on Labs? 46:21 Multimodal Is Still Missing Something 49:46 OpenAI's Bet on Reasoning 55:26 The AI Coding Wars 59:26 Focus vs. Keeping Embers Burning 1:02:09 Open Source vs. Closed Source Gap 1:05:15 Quickfire YouTube: Spotify: Apple:
1:13:33

I sat down with Lukasz Kaiser to get into whether the architecture he helped invent is actually enough, and what's next in generalization, coding agents, RL and more. Lukasz co-authored "Attention Is All You Need," the paper that introduced the transformer and worked on reasoning models at OpenAI so he’s been a key part of major shifts in the field. We hit on: ▪️ The case for and against a new architecture coming after the transformer ▪️ What’s required for model generalization in the physical world ▪️ How much coding agents have improved his AI research productivity ▪️ The next domains for RL ▪️ Why Anthropic initially won coding ▪️ Future research directions he’s excited about 0:00 Intro 1:12 Transformers vs. Human Learning 8:37 How Do We Get Physical World Generalization? 10:52 What Comes After Transformers 13:59 How Much Have Agents Improved Lukasz's AI Research Productivity? 17:21 How Close Is an AI Research Intern? 26:06 RL Beyond Verifiable Tasks 35:38 App Companies: Build Models or Lean on Labs? 46:21 Multimodal Is Still Missing Something 49:46 OpenAI's Bet on Reasoning 55:26 The AI Coding Wars 59:26 Focus vs. Keeping Embers Burning 1:02:09 Open Source vs. Closed Source Gap 1:05:15 Quickfire YouTube: Spotify: Apple:

Jacob Effron

75,036 Aufrufe • vor 1 Monat

OpenAI CEO Sam Altman predicts that we could achieve AI superintelligence by 2025. He highlighted the potential of AI agents to handle complex tasks independently, much like a skilled human. While the progress in deep learning has been groundbreaking, he emphasized that there’s still a lot of hard work ahead to ensure these advancements benefit humanity responsibly.
1:49

OpenAI CEO Sam Altman predicts that we could achieve AI superintelligence by 2025. He highlighted the potential of AI agents to handle complex tasks independently, much like a skilled human. While the progress in deep learning has been groundbreaking, he emphasized that there’s still a lot of hard work ahead to ensure these advancements benefit humanity responsibly.

Andy Parackal

28,768 Aufrufe • vor 1 Jahr

Talked all things AI with Swaroop Mishra, Research Scientist at Google DeepMind 🧠, diving into his work on Gemini and what the shift to AI-assisted coding actually means for software engineers. We covered: - Building Bard & Gemini - Early days of instruction tuning (2021) - The shift from writing code to reviewing it - The future of token usage and measuring ROI - Why first principles still matter most Checkout the full video 👇
14:53

Talked all things AI with Swaroop Mishra, Research Scientist at Google DeepMind 🧠, diving into his work on Gemini and what the shift to AI-assisted coding actually means for software engineers. We covered: - Building Bard & Gemini - Early days of instruction tuning (2021) - The shift from writing code to reviewing it - The future of token usage and measuring ROI - Why first principles still matter most Checkout the full video 👇

Khushboo Verma

24,855 Aufrufe • vor 1 Monat

Capability overhang means too many gaps today between what the models can do and what most people actually do with them. 2026 Prediction: Progress towards AGI will depend as much on helping people use AI well, in ways that directly benefit them as on progress in frontier models themselves. 2026 will be about frontier research AND about closing this deployment gap — especially in health care, business, and people's daily lives.
1:02

Capability overhang means too many gaps today between what the models can do and what most people actually do with them. 2026 Prediction: Progress towards AGI will depend as much on helping people use AI well, in ways that directly benefit them as on progress in frontier models themselves. 2026 will be about frontier research AND about closing this deployment gap — especially in health care, business, and people's daily lives.

OpenAI

666,839 Aufrufe • vor 7 Monaten

In this episode, OpenAI Chief Research Officer Mark Chen joins allen to flambé shrimp, cook Korean stew, and chat about being at the frontier of AI research: why scaling laws and pre-training still matter, how OpenAI chooses research bets and allocates compute, what it means to develop research taste, why evals are in crisis, how to avoid benchmark-maxing, and what it will take for models to handle long-horizon real-world work, multimodal reasoning, and eventually end-to-end AI research. Timestamps: 0:00 Intro 0:28 The Soup Story 1:52 From Trading to AI Research 3:21 How to Develop Research Taste 5:23 RL, Evals, and Superhuman Benchmarks 8:17 Cooking Begins on the Impulse Stove 8:53 Scaling Laws, Pre-Training, and Reasoning 12:33 OpenAI’s Research Roadmap and Compute Allocation 15:48 What Makes a Great Researcher 19:33 The Evals Crisis and Benchmark-Maxing 24:34 Jagged Intelligence, Context, and Long-Horizon Learning 27:14 Shrimp Flambé and New Research Bets 31:32 Multimodal Models and One Architecture 32:36 Vibe Researching and End-to-End AI Research 34:36 Failed Bets, Postmortems, and OpenAI’s Alpha 37:07 Final Taste Test 37:53 Overrated vs. Underrated AI Research 41:00 Closing
41:16

In this episode, OpenAI Chief Research Officer Mark Chen joins allen to flambé shrimp, cook Korean stew, and chat about being at the frontier of AI research: why scaling laws and pre-training still matter, how OpenAI chooses research bets and allocates compute, what it means to develop research taste, why evals are in crisis, how to avoid benchmark-maxing, and what it will take for models to handle long-horizon real-world work, multimodal reasoning, and eventually end-to-end AI research. Timestamps: 0:00 Intro 0:28 The Soup Story 1:52 From Trading to AI Research 3:21 How to Develop Research Taste 5:23 RL, Evals, and Superhuman Benchmarks 8:17 Cooking Begins on the Impulse Stove 8:53 Scaling Laws, Pre-Training, and Reasoning 12:33 OpenAI’s Research Roadmap and Compute Allocation 15:48 What Makes a Great Researcher 19:33 The Evals Crisis and Benchmark-Maxing 24:34 Jagged Intelligence, Context, and Long-Horizon Learning 27:14 Shrimp Flambé and New Research Bets 31:32 Multimodal Models and One Architecture 32:36 Vibe Researching and End-to-End AI Research 34:36 Failed Bets, Postmortems, and OpenAI’s Alpha 37:07 Final Taste Test 37:53 Overrated vs. Underrated AI Research 41:00 Closing

Latent.Space

62,533 Aufrufe • vor 1 Monat

Leopold Aschenbrenner, the 24 year old who wrote a 165 page AGI manifesto, got it right on the money, and turned it into a $5.5 billion hedge fund. And he's identifying the single most important milestone to watch for in all of AI. The question is can AI automate AI research itself? Here's why that question matters so much. Right now, a few thousand human researchers at the frontier labs are driving all the progress. They design experiments, write papers, propose architectural improvements, build the next generation of models and it's an incredibly small workforce doing incredibly high-leverage work. If an AI system can do that job even partially, the feedback loop changes completely. The AI makes algorithmic improvements, which produces more powerful AI, which makes better improvements, faster. You go from linear progress to compounding returns and a decade of research could compress into a year. Aschenbrenner says there's a "pretty reasonable chance" this happens within five years. He's not alone, Anthropic says they're on track to fully automate AI R&D as soon as early 2027. OpenAI has publicly targeted a fully autonomous AI researcher by March 2028 and Sam Altman has said a research intern level AI will exist before the end of this year. If he's right, the next few years won't look like the last few years but they'll look like nothing we've seen before. The future is bright!
0:49

Leopold Aschenbrenner, the 24 year old who wrote a 165 page AGI manifesto, got it right on the money, and turned it into a $5.5 billion hedge fund. And he's identifying the single most important milestone to watch for in all of AI. The question is can AI automate AI research itself? Here's why that question matters so much. Right now, a few thousand human researchers at the frontier labs are driving all the progress. They design experiments, write papers, propose architectural improvements, build the next generation of models and it's an incredibly small workforce doing incredibly high-leverage work. If an AI system can do that job even partially, the feedback loop changes completely. The AI makes algorithmic improvements, which produces more powerful AI, which makes better improvements, faster. You go from linear progress to compounding returns and a decade of research could compress into a year. Aschenbrenner says there's a "pretty reasonable chance" this happens within five years. He's not alone, Anthropic says they're on track to fully automate AI R&D as soon as early 2027. OpenAI has publicly targeted a fully autonomous AI researcher by March 2028 and Sam Altman has said a research intern level AI will exist before the end of this year. If he's right, the next few years won't look like the last few years but they'll look like nothing we've seen before. The future is bright!

Milk Road AI

607,502 Aufrufe • vor 2 Monaten

Will SaaS survive the AI era? Four reasons it may not: - Complexity: Software will be cheaper & easier to build yourself - Model Providers: Foundation model providers will move into the app layer. - Agentic Models: Models are becoming more agentic, they could handle tasks without extra SaaS layers. - Experience: SaaS today is built for human GUIs, AI in the future would not require any clicking. Link to the entire interview with Harry Stebbings below.
1:04

Will SaaS survive the AI era? Four reasons it may not: - Complexity: Software will be cheaper & easier to build yourself - Model Providers: Foundation model providers will move into the app layer. - Agentic Models: Models are becoming more agentic, they could handle tasks without extra SaaS layers. - Experience: SaaS today is built for human GUIs, AI in the future would not require any clicking. Link to the entire interview with Harry Stebbings below.

Turing

34,425 Aufrufe • vor 7 Monaten

.Jerry Tworek helped drive o1, o3, and Codex at OpenAI where he was VP of Research from 2019 to 2025. Then he left to pursue “types of research that are hard to do at OpenAI.” This week on Unsupervised Learning, I sat down with Jerry to discuss where AI research is headed and what he learned from seven years at the forefront of the field. - Why he left OpenAI after helping create some of its biggest breakthroughs - Why Jerry updated his AGI timeline after building reasoning models - The real limits of scaling reinforcement learning - Why Anthropic has done so well in coding - Inside OpenAI's pivotal decisions - What makes great AI researchers Timestamps: 0:00 Intro 1:26 Scaling Paradigms in AI 3:36 Challenges in Reinforcement Learning 11:48 AGI Timelines 18:36 Converging Labs and Economic Forces 25:05 Jerry's Departure from OpenAI 31:18 Pivotal Decisions in OpenAI's Journey 35:06 Balancing Research and Product Development 38:42 The Future of AI Coding 41:33 Specialization vs. Generalization in AI 48:47 Hiring and Building Research Teams 55:21 Quickfire Listen here: YouTube: Spotify: Apple:
1:02:59

.Jerry Tworek helped drive o1, o3, and Codex at OpenAI where he was VP of Research from 2019 to 2025. Then he left to pursue “types of research that are hard to do at OpenAI.” This week on Unsupervised Learning, I sat down with Jerry to discuss where AI research is headed and what he learned from seven years at the forefront of the field. - Why he left OpenAI after helping create some of its biggest breakthroughs - Why Jerry updated his AGI timeline after building reasoning models - The real limits of scaling reinforcement learning - Why Anthropic has done so well in coding - Inside OpenAI's pivotal decisions - What makes great AI researchers Timestamps: 0:00 Intro 1:26 Scaling Paradigms in AI 3:36 Challenges in Reinforcement Learning 11:48 AGI Timelines 18:36 Converging Labs and Economic Forces 25:05 Jerry's Departure from OpenAI 31:18 Pivotal Decisions in OpenAI's Journey 35:06 Balancing Research and Product Development 38:42 The Future of AI Coding 41:33 Specialization vs. Generalization in AI 48:47 Hiring and Building Research Teams 55:21 Quickfire Listen here: YouTube: Spotify: Apple:

Jacob Effron

43,788 Aufrufe • vor 5 Monaten

We benchmarked leading multimodal foundation models (GPT-4o, Claude 3.5 Sonnet, Gemini, Llama, etc.) on standard computer vision tasks—from segmentation to surface normal estimation—using standard datasets like COCO and ImageNet. These models have made remarkable progress; however, it is unclear exactly where they stand in terms of understanding vision in detail. Especially when it comes to tasks beyond question-answering. How well do they understand an object's segments or geometry? Our analyses yield an assessment that is quantitatively and qualitatively detailed and is compatible with evaluations developed in the field of computer vision over the past decades. Observed trends: 🔹 The foundation models consistently underperform task-specific SOTA models across all tasks. However, they are respectable generalists, which is remarkable as they are presumably trained primarily on image-text-based tasks. 🔹 They perform semantic tasks notably better than geometric ones. 🔹 GPT-4o performs the best among non-reasoning models, getting the top position in 4 out of 6 tasks. 🔹 Reasoning models, e.g., o3, show improvements in geometric tasks. 🔹 The 'image generation' models, e.g., GPT-40 Image Generation, which have been natively trained multimodally, exhibit quirks. E.g., hallucinated objects, misalignment between the input and output, etc. 🔹 While the prompting techniques affect performance, better models exhibit less sensitivity to variations in prompts. We control for the variance introduced by the prompting methods in our experiments. 🌐 Detailed analyses, visualizations: ⌨️ code: 🧵 1/n
2:36

We benchmarked leading multimodal foundation models (GPT-4o, Claude 3.5 Sonnet, Gemini, Llama, etc.) on standard computer vision tasks—from segmentation to surface normal estimation—using standard datasets like COCO and ImageNet. These models have made remarkable progress; however, it is unclear exactly where they stand in terms of understanding vision in detail. Especially when it comes to tasks beyond question-answering. How well do they understand an object's segments or geometry? Our analyses yield an assessment that is quantitatively and qualitatively detailed and is compatible with evaluations developed in the field of computer vision over the past decades. Observed trends: 🔹 The foundation models consistently underperform task-specific SOTA models across all tasks. However, they are respectable generalists, which is remarkable as they are presumably trained primarily on image-text-based tasks. 🔹 They perform semantic tasks notably better than geometric ones. 🔹 GPT-4o performs the best among non-reasoning models, getting the top position in 4 out of 6 tasks. 🔹 Reasoning models, e.g., o3, show improvements in geometric tasks. 🔹 The 'image generation' models, e.g., GPT-40 Image Generation, which have been natively trained multimodally, exhibit quirks. E.g., hallucinated objects, misalignment between the input and output, etc. 🔹 While the prompting techniques affect performance, better models exhibit less sensitivity to variations in prompts. We control for the variance introduced by the prompting methods in our experiments. 🌐 Detailed analyses, visualizations: ⌨️ code: 🧵 1/n

Amir Zamir

73,074 Aufrufe • vor 1 Jahr

Databricks is excited to partner with OpenAI on GPT-5.5, their latest frontier model. GPT-5.5 will be available in Unity AI Gateway on launch. You can use it with coding tools such as Codex, or to power your enterprise agents. GPT-5.5 is state-of-the-art on many benchmarks including OfficeQA Pro, our benchmark for evaluating grounded reasoning on enterprise tasks. We are partnering with OpenAI to co-launch on Databricks. Hear more from our co-founder Patrick Wendell and OpenAI CRO Denise Holland Dresser on GPT-5.5 in Databricks.
1:53

Databricks is excited to partner with OpenAI on GPT-5.5, their latest frontier model. GPT-5.5 will be available in Unity AI Gateway on launch. You can use it with coding tools such as Codex, or to power your enterprise agents. GPT-5.5 is state-of-the-art on many benchmarks including OfficeQA Pro, our benchmark for evaluating grounded reasoning on enterprise tasks. We are partnering with OpenAI to co-launch on Databricks. Hear more from our co-founder Patrick Wendell and OpenAI CRO Denise Holland Dresser on GPT-5.5 in Databricks.

Databricks

12,707 Aufrufe • vor 3 Monaten

Today we're launching Scholara, the world's first AI-native systematic review platform. The systematic review is the gold standard of evidence in research. It's also one of the most time-consuming and costly things a researcher will ever do with an average of 68 weeks to complete. We rebuilt the entire pipeline into a sub 2-hour conversation. Scholara is already used and trusted by thousands of researchers, completing systematic reviews in hours.
3:30

Today we're launching Scholara, the world's first AI-native systematic review platform. The systematic review is the gold standard of evidence in research. It's also one of the most time-consuming and costly things a researcher will ever do with an average of 68 weeks to complete. We rebuilt the entire pipeline into a sub 2-hour conversation. Scholara is already used and trusted by thousands of researchers, completing systematic reviews in hours.

Liam

21,400 Aufrufe • vor 1 Monat

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
58:45

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 Aufrufe • vor 3 Monaten

Demis Hassabis recommendation for college students. He’s still do STEM, math and computer science. Expertise in those fields will help better leverage AI for at least next decade. Those in non-techical majors, really “lean in to” using latest models. AI labs spending so much time creating new models that they’ve only “scratched the surface” of what the models can actually do (huge “capability overhang”). And expertise in any field can be turbocharged by smart use of AI. “Double down on your own agency. The future is still to be written. Don’t listen to anyone that says it’s not.”
3:00

Demis Hassabis recommendation for college students. He’s still do STEM, math and computer science. Expertise in those fields will help better leverage AI for at least next decade. Those in non-techical majors, really “lean in to” using latest models. AI labs spending so much time creating new models that they’ve only “scratched the surface” of what the models can actually do (huge “capability overhang”). And expertise in any field can be turbocharged by smart use of AI. “Double down on your own agency. The future is still to be written. Don’t listen to anyone that says it’s not.”

Bearly AI

218,036 Aufrufe • vor 1 Monat

Epiplexity, Reasoning & The "Alien" Behavior of LLMs: my conversation with Pavel Izmailov, AI researcher at Anthropic and professor at New York University. 00:00 - Intro 00:53 - Alien survival instincts: is continual learning about to create a major alignment problem? Riffing on the post by 🍓🍓🍓 03:33 - Did AI learn deception from sci-fi literature? 05:55 - Defining Alignment, Superalignment & OpenAI teams 08:12 - Pavel’s journey: From Russian math to OpenAI Superalignment 10:46 - Culture check: OpenAI vs. Anthropic vs. Academia 11:54 - Why move to NYU? Academia and the need for exploratory research 13:09 - Does reasoning make AI alignment harder or easier? 14:22 - Sandbagging: When models pretend to be dumb 16:19 - Scalable Oversight: Using AI to supervise AI 18:04 - Weak-to-Strong Generalization 22:43 - Mechanistic Interpretability: Inside the black box 25:08 - The reasoning explosion 27:07 - Are Transformers enough or do we need a new paradigm? 28:29 - RL vs. Test-Time Compute: What’s actually driving progress? 30:10 - Long-horizon tasks: Agents running for hours 31:49 - Epiplexity: A new theory of data information content - Pavel's work with Marc Finzi, Zico Kolter, Andrew Gordon Wilson, Yiding Jiang and Shikai Qiu. 38:29 - 2026 Predictions: Multi-agent systems & reasoning limits 39:28 - AI & Scientific progress 41:42 - Advice for PhD students
45:01

Epiplexity, Reasoning & The "Alien" Behavior of LLMs: my conversation with Pavel Izmailov, AI researcher at Anthropic and professor at New York University. 00:00 - Intro 00:53 - Alien survival instincts: is continual learning about to create a major alignment problem? Riffing on the post by 🍓🍓🍓 03:33 - Did AI learn deception from sci-fi literature? 05:55 - Defining Alignment, Superalignment & OpenAI teams 08:12 - Pavel’s journey: From Russian math to OpenAI Superalignment 10:46 - Culture check: OpenAI vs. Anthropic vs. Academia 11:54 - Why move to NYU? Academia and the need for exploratory research 13:09 - Does reasoning make AI alignment harder or easier? 14:22 - Sandbagging: When models pretend to be dumb 16:19 - Scalable Oversight: Using AI to supervise AI 18:04 - Weak-to-Strong Generalization 22:43 - Mechanistic Interpretability: Inside the black box 25:08 - The reasoning explosion 27:07 - Are Transformers enough or do we need a new paradigm? 28:29 - RL vs. Test-Time Compute: What’s actually driving progress? 30:10 - Long-horizon tasks: Agents running for hours 31:49 - Epiplexity: A new theory of data information content - Pavel's work with Marc Finzi, Zico Kolter, Andrew Gordon Wilson, Yiding Jiang and Shikai Qiu. 38:29 - 2026 Predictions: Multi-agent systems & reasoning limits 39:28 - AI & Scientific progress 41:42 - Advice for PhD students

Matt Turck

23,020 Aufrufe • vor 6 Monaten

Thanksgiving-week treat: an epic conversation on Frontier AI with Lukasz Kaiser -co-author of “Attention Is All You Need” (Transformers) and leading research scientist at OpenAI working on GPT-5.1-era reasoning models. 00:00 – Cold open and intro 01:29 – “AI slowdown” vs a wild week of new frontier models 08:03 – Low-hanging fruit, infra, RL training and better data 11:39 – What is a reasoning model, in plain language 17:02 – Chain-of-thought and training the thinking process with RL 21:39 – Łukasz’s path: from logic and France to Google and Kurzweil 24:20 – Inside the Transformer story and what “attention” really means 28:42 – From Google Brain to OpenAI: culture, scale and GPUs 32:49 – What’s next for pre-training, GPUs and distillation 37:29 – Can we still understand these models? Circuits, sparsity and black boxes 39:42 – GPT-4 → GPT-5 → GPT-5.1: what actually changed 42:40 – Post-training, safety and teaching GPT-5.1 different tones 46:16 – How long should GPT-5.1 think? Reasoning tokens and jagged abilities 47:43 – The five-year-old’s dot puzzle that still breaks frontier models 52:22 – Generalization, child-like learning and whether reasoning is enough 53:48 – Beyond Transformers: ARC, LeCun’s ideas and multimodal bottlenecks 56:10 – GPT-5.1 Codex Max, long-running agents and compaction 1:00:06 – Will foundation models eat most apps? The translation analogy and trust 1:02:34 – What still needs to be solved, and where AI might go next
1:05:25

Thanksgiving-week treat: an epic conversation on Frontier AI with Lukasz Kaiser -co-author of “Attention Is All You Need” (Transformers) and leading research scientist at OpenAI working on GPT-5.1-era reasoning models. 00:00 – Cold open and intro 01:29 – “AI slowdown” vs a wild week of new frontier models 08:03 – Low-hanging fruit, infra, RL training and better data 11:39 – What is a reasoning model, in plain language 17:02 – Chain-of-thought and training the thinking process with RL 21:39 – Łukasz’s path: from logic and France to Google and Kurzweil 24:20 – Inside the Transformer story and what “attention” really means 28:42 – From Google Brain to OpenAI: culture, scale and GPUs 32:49 – What’s next for pre-training, GPUs and distillation 37:29 – Can we still understand these models? Circuits, sparsity and black boxes 39:42 – GPT-4 → GPT-5 → GPT-5.1: what actually changed 42:40 – Post-training, safety and teaching GPT-5.1 different tones 46:16 – How long should GPT-5.1 think? Reasoning tokens and jagged abilities 47:43 – The five-year-old’s dot puzzle that still breaks frontier models 52:22 – Generalization, child-like learning and whether reasoning is enough 53:48 – Beyond Transformers: ARC, LeCun’s ideas and multimodal bottlenecks 56:10 – GPT-5.1 Codex Max, long-running agents and compaction 1:00:06 – Will foundation models eat most apps? The translation analogy and trust 1:02:34 – What still needs to be solved, and where AI might go next

Matt Turck

168,007 Aufrufe • vor 8 Monaten

🚀 AI Just Got a Major Upgrade! Say hello to Baidu ERNIE’s new reasoning model — built to push the limits of what AI can do. ✨ Sharper Factual Accuracy – delivers reliable, verified knowledge. ✨ Smarter Instruction Following – understands complex tasks with ease. ✨ Next-Level Reasoning Power – excels in coding, logic, and creative writing. Powered by the ERNIE-4.5 base and reinforced through advanced multi-round reinforcement learning, this model doesn’t just follow commands — it thinks, plans, and executes like never before. From solving real-world problems to generating innovative ideas, Wenxin Yiyan is redefining the AI frontier. 🔥 The future of AI is here →
2:34

🚀 AI Just Got a Major Upgrade! Say hello to Baidu ERNIE’s new reasoning model — built to push the limits of what AI can do. ✨ Sharper Factual Accuracy – delivers reliable, verified knowledge. ✨ Smarter Instruction Following – understands complex tasks with ease. ✨ Next-Level Reasoning Power – excels in coding, logic, and creative writing. Powered by the ERNIE-4.5 base and reinforced through advanced multi-round reinforcement learning, this model doesn’t just follow commands — it thinks, plans, and executes like never before. From solving real-world problems to generating innovative ideas, Wenxin Yiyan is redefining the AI frontier. 🔥 The future of AI is here →

Toha Khan

32,668 Aufrufe • vor 10 Monaten

Why does AI suck at design? I asked OpenAI Codex lead Andrew Ambrosino His take: 1. Practically, labs prioritized coding because it accelerates AI research, and good design is harder to grade than good code. 2. Structurally, great design requires novelty and an understanding of the culture. A model that outputs the Linear website every time isn’t good design. Models currently index heavily toward known patterns vs. something new. He believes the practical issues will be solved, but the issues around novelty and culture may take a while.
0:55

Why does AI suck at design? I asked OpenAI Codex lead Andrew Ambrosino His take: 1. Practically, labs prioritized coding because it accelerates AI research, and good design is harder to grade than good code. 2. Structurally, great design requires novelty and an understanding of the culture. A model that outputs the Linear website every time isn’t good design. Models currently index heavily toward known patterns vs. something new. He believes the practical issues will be solved, but the issues around novelty and culture may take a while.

Lenny Rachitsky

40,764 Aufrufe • vor 24 Tagen

Sora 2 is here. I’ve been pushing its limits to see how far it can really go. Some results look impressive, others still rough but the biggest plus is that now it can generate video with audio in one go. 🔥 The downside? The watermark is still everywhere, and the raw video quality isn’t great yet. For this demo I had to upscale with Topaz Labs Clearly the devs still have work to do. But the progress is undeniable, smoother motion, stronger scene understanding and it’s beginning to handle even the most complex prompts I throw at it, turning them into fully visualized videos. Still early days but definitely exciting to watch. Enjoy the show✨ OpenAI #Sora2
1:15

Sora 2 is here. I’ve been pushing its limits to see how far it can really go. Some results look impressive, others still rough but the biggest plus is that now it can generate video with audio in one go. 🔥 The downside? The watermark is still everywhere, and the raw video quality isn’t great yet. For this demo I had to upscale with Topaz Labs Clearly the devs still have work to do. But the progress is undeniable, smoother motion, stronger scene understanding and it’s beginning to handle even the most complex prompts I throw at it, turning them into fully visualized videos. Still early days but definitely exciting to watch. Enjoy the show✨ OpenAI #Sora2

DStudioproject

50,387 Aufrufe • vor 9 Monaten