François Chollet

François Chollet has spent years asking a different question than most of the AI world. Instead of scaling what already works, he’s trying to understand what intelligence actually is—and how to build it from first principles. In this episode of Lightcone, he traces that path from his early work on deep learning to the creation of the ARC prize, and the launch of ARC V3, a new benchmark designed to measure something deeper than performance: the ability to learn, adapt, and reason efficiently in entirely new environments. He explains why today’s systems may be hitting limits, what recent breakthroughs really mean, and why reaching true general intelligence may require a fundamentally different approach.00:00 - AGI by 2030?00:31 - Introducing Ndea: A New Path Beyond Deep Learning01:08 - A New ML Paradigm 01:30 - Replacing neural nets with compact symbolic programs03:04 - Why Ndea Isn’t Competing With Coding Agents05:20 - Why Everyone Might Be Wrong About Scaling LLMs07:22 - Why Coding Agents Suddenly Work So Well08:50 - The Limits of LLMs in Non-Verifiable Domains10:48 - What AGI Actually Means (And Why Most Definitions Are Wrong)13:30 - Why Deep Learning Hits a Wall 14:00 - ARC’s Origin Story18:20 - ARC Benchmarks Explained: From V1 to V322:49 - The RL Loop Powering Coding Agents Today27:03 - ARC-AGI V3: Measuring “Agentic Intelligence”31:14 - Inside the ARC Game Studio35:31 - Could AGI Fit in 10,000 Lines of Code?44:01 - Building Ndea: From Idea to Compounding Research Stack46:46 - The Future of ARC: Benchmarks That Evolve With AI47:21 - Why There’s Still Huge Opportunity for New AI Paradigms53:37 - How to Build a Breakout Open Source Project - Lessons From Kera56:39 - Advice For How To Think About AIApply to Y Combinator: https://www.ycombinator.com/applyWork at a startup: https://www.ycombinator.com/jobs

OpenAI invierte 94 millones en Isara, una startup de coordinación de agentes de IA a escala masiva fundada por dos veinteañeros. François Chollet lanza ARC-AGI-3, un benchmark interactivo con videojuegos que mide eficiencia de aprendizaje real. El objeto interestelar 3I/ATLAS muestra concentraciones de deuterio diez veces mayores que cualquier cometa conocido — la explicación apunta a su formación hace 12.000 millones de años en condiciones extremas. GitHub Copilot usará tu código como dato de entrenamiento a partir del 24 de abril en modo opt-out. Y Bernie Sanders presenta una moratoria indefinida sobre centros de datos de IA que superen los 20 megavatios, conectando la infraestructura tech con el precio de la electricidad. Puedes seguirnos en YouTube en https://youtube.com/olivernabani y puedes unirte al Discord Mashain en https://olivernabani.com/discord

作者François Chollet (弗朗索瓦·肖莱) 是一位在人工智能领域极具影响力的法国研究员。他的核心成就与贡献是开发了Keras，一个开源的深度学习框架，Keras以其用户友好、模块化和可扩展性，极大地降低了深度学习的入门门槛，迅速成为全球最受欢迎的深度学习库之一。在2015年发布Keras后不久，Chollet加入了Google长期担任高级软件工程师，继续领导Keras的开发，并对TensorFlow生态系统做出了重大贡献。除了工程上的巨大成功，Chollet更是一位深入思考人工智能本质的研究者。在《关于智能的衡量》(On the Measure of Intelligence)这篇影响深远的论文中，他提出了一个全新的智能定义：“智能是系统在面对新任务时，基于先验知识和经验，高效获取新技能的能力”。这个定义强调的是学习效率和泛化能力，而非在单一任务上的表现。为了实践他的智能理论... 去小宇宙查看完整单集简介 在小宇宙查看该单集文稿

Google's Video AI Veo 3 Tested: Costs, Features, Limits. Salesforce and Co.: How Much Work Does AI Already Take Over – and What Does That Mean?. François Chollet on the End of Scaling, ARC-3, and His Path to AGI. Robotic Probe Quickly Measures Key Properties of New Materials. The AI news for July 5th, 2025 Here are the details of the day's selected top stories: News #1: Google's Video AI Veo 3 Tested: Costs, Features, Limits Source: https://www.heise.de/tests/Googles-Video-KI-Veo-3-ausprobiert-Kosten-Funktionen-Limits-10474831.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag_plus.beitrag_plus Why did we choose this article? This article provides an in-depth look at Google's new video AI technology, Veo 3, which offers photorealistic videos with sound. It discusses the costs, features, and limitations of the service, making it a valuable read for those interested in the latest AI advancements in media technology. News #2: Salesforce and Co.: How Much Work Does AI Already Take Over – and What Does That Mean? Source: https://www.heise.de/tests/Salesforce-und-Co-Wie-viel-Arbeit-uebernimmt-die-KI-schon-und-was-heisst-das-10475379.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag Why did we choose this article? This article explores the impact of AI on the workforce, particularly in companies like Google, Salesforce, and Microsoft. It raises important questions about the extent of AI's role in automating tasks and its implications for employees, offering a balanced view on the evolving workplace. News #3: François Chollet on the End of Scaling, ARC-3, and His Path to AGI Source: https://the-decoder.de/francois-chollet-ueber-das-ende-der-skalierung-arc-3-und-seinen-weg-zu-agi/ Why did we choose this article? This article features insights from AI researcher François Chollet, who critiques the current trend of scaling AI models and proposes a new approach to developing AGI. His perspective on adaptive learning and problem-solving offers a fresh take on the future of AI development. News #4: Robotic Probe Quickly Measures Key Properties of New MaterialsSource: https://news.mit.edu/2025/robotic-probe-quickly-measures-key-properties-new-materials-0704 Why did we choose this article? This article highlights a breakthrough in material science with the development of a robotic probe that can quickly analyze new semiconductors. This innovation could significantly impact the creation of more efficient solar cells, showcasing AI's potential in advancing sustainable technologies. Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at podcast@pickert.de. Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Neues KI-Modell sagt menschliche Entscheidungen vorher. Europas KI-Pläne auf Eis? Konzerne fordern Stop für AI Act. KI jagt Verkehrssünder in Echtzeit – Athen geht voran. François Chollet über das Ende der Skalierung, ARC-3 und seinen Weg zu AGI. Die KI-News vom 05.07.2025. Hier die Details zu den ausgewählten News des Tages: 1. Nachricht: Neues KI-Modell sagt menschliche Entscheidungen vorher Quelle: https://www.all-ai.de/news/topbeitraege/ki-analyse-entscheidungen Warum haben wir diesen Artikel ausgewählt? Dieser Artikel ist besonders informativ, da er ein neues KI-Modell vorstellt, das menschliche Entscheidungen präzise vorhersagen kann. Die Verbindung von KI und Psychologie ist ein spannendes Thema, das sowohl methodische Fortschritte als auch ethische Fragen aufwirft. Die Möglichkeit, menschliches Verhalten besser zu verstehen und vorherzusagen, bietet großes Potenzial für die Forschung, birgt jedoch auch Risiken, die diskutiert werden müssen. 2. Nachricht: Europas KI-Pläne auf Eis? Konzerne fordern Stop für AI Act Quelle: https://www.all-ai.de/news/news24/europa-kiact-stopp Warum haben wir diesen Artikel ausgewählt? Die Diskussion um den AI Act ist hochaktuell und relevant, da sie die Balance zwischen Innovation und Regulierung in Europa betrifft. Der Artikel beleuchtet die Forderungen großer Unternehmen nach einem Aufschub der Gesetzgebung und die möglichen Auswirkungen auf die Innovationsfähigkeit. Diese Debatte ist entscheidend für die zukünftige Entwicklung von KI in Europa. 3. Nachricht: KI jagt Verkehrssünder in Echtzeit – Athen geht voran Quelle: https://www.all-ai.de/news/news24/ki-strafzettel-athen Warum haben wir diesen Artikel ausgewählt? Dieser Artikel ist interessant, da er ein praktisches Anwendungsbeispiel von KI im Bereich der Verkehrsüberwachung beschreibt. Das Pilotprojekt in Athen zeigt, wie KI zur Effizienzsteigerung und zur Einhaltung von Verkehrsregeln beitragen kann, während gleichzeitig Datenschutzaspekte berücksichtigt werden. Solche Anwendungen könnten auch in anderen Städten relevant werden. 4. Nachricht: François Chollet über das Ende der Skalierung, ARC-3 und seinen Weg zu AGI Quelle: https://the-decoder.de/francois-chollet-ueber-das-ende-der-skalierung-arc-3-und-seinen-weg-zu-agi/ Warum haben wir diesen Artikel ausgewählt? Der Artikel bietet eine einzigartige Perspektive auf die zukünftige Entwicklung von KI durch die Augen eines renommierten Forschers. François Chollet diskutiert das Ende der Skalierung großer Modelle und skizziert eine neue Ära der KI-Entwicklung. Diese Einsichten sind wertvoll für das Verständnis der strategischen Ausrichtung der KI-Forschung und ihrer potenziellen Auswirkungen. Hast du Fragen, Kommentare oder Verbesserungsvorschläge? Wir freuen über Feedback an podcast@pickert.de. Möchtest du selbst einen solchen KI-generierten und 100% automatisierten Podcast zu deinem Thema haben? --> Melde dich bei uns, wir machen es möglich.

François Chollet on June 16, 2025 at AI Startup School in San Francisco. François Chollet is a leading voice in AI. He's the creator of the Keras library, author of Deep Learning with Python, and the founder of the ARC Prize, a global competition aimed at measuring true general intelligence. He's spent years thinking deeply about what intelligence actually is—and why scaling up today’s AI models isn’t enough to reach it. In this talk, he walks through the limits of pretraining and memorized skills, and lays out a path toward true general intelligence—AI that can adapt on the fly, reason in new situations, and invent novel solutions. He explains why abstraction and compositionality matter, how ARC became the benchmark for progress, and what his team at a new research lab called Ndea is building next.

In this fascinating episode, we dive deep into the race towards true AI intelligence, AGI benchmarks, test-time adaptation, and program synthesis with star AI researcher (and philosopher) Francois Chollet, creator of Keras and the ARC AGI benchmark, and Mike Knoop, co-founder of Zapier and now co-founder with Francois of both the ARC Prize and the research lab Ndea. With the launch of ARC Prize 2025 and ARC-AGI 2, they explain why existing LLMs fall short on true intelligence tests, how new models like O3 mark a step change in capabilities, and what it will really take to reach AGI. We cover everything from the technical evolution of ARC 1 to ARC 2, the shift toward test-time reasoning, and the role of program synthesis as a foundation for more general intelligence. The conversation also explores the philosophical underpinnings of intelligence, the structure of the ARC Prize, and the motivation behind launching Ndea — a ew AGI research lab that aims to build a "factory for rapid scientific advancement." Whether you're deep in the AI research trenches or just fascinated by where this is all headed, this episode offers clarity and inspiration. Ndea Website - https://ndea.com X/Twitter - https://x.com/ndea ARC Prize Website - https://arcprize.org X/Twitter - https://x.com/arcprize François Chollet LinkedIn - https://www.linkedin.com/in/fchollet X/Twitter - https://x.com/fchollet Mike Knoop X/Twitter - https://x.com/mikeknoop FIRSTMARK Website - https://firstmark.com X/Twitter - https://twitter.com/FirstMarkCap Matt Turck (Managing Director) LinkedIn - https://www.linkedin.com/in/turck/ X/Twitter - https://twitter.com/mattturck (00:00) Intro (01:05) Introduction to ARC Prize 2025 and ARC-AGI 2 (02:07) What is ARC and how it differs from other AI benchmarks (02:54) Why current models struggle with fluid intelligence (03:52) Shift from static LLMs to test-time adaptation (04:19) What ARC measures vs. traditional benchmarks (07:52) Limitations of brute-force scaling in LLMs (13:31) Defining intelligence: adaptation and efficiency (16:19) How O3 achieved a massive leap in ARC performance (20:35) Speculation on O3's architecture and test-time search (22:48) Program synthesis: what it is and why it matters (28:28) Combining LLMs with search and synthesis techniques (34:57) The ARC Prize structure: efficiency track, private vs. public (42:03) Open source as a requirement for progress (44:59) What's new in ARC-AGI 2 and human benchmark testing (48:14) Capabilities ARC-AGI 2 is designed to test (49:21) When will ARC-AGI 2 be saturated? AGI timelines (52:25) Founding of NDEA and why now (54:19) Vision beyond AGI: a factory for scientific advancement (56:40) What NDEA is building and why it's different from LLM labs (58:32) Hiring and remote-first culture at NDEA (59:52) Closing thoughts and the future of AI research

We are joined by Francois Chollet and Mike Knoop, to launch the new version of the ARC prize! In version 2, the challenges have been calibrated with humans such that at least 2 humans could solve each task in a reasonable task, but also adversarially selected so that frontier reasoning models can't solve them. The best LLMs today get negligible performance on this challenge. https://arcprize.org/ SPONSOR MESSAGES: *** Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. They are hiring a Chief Engineer and ML engineers. Events in Zurich. Goto https://tufalabs.ai/ *** TRANSCRIPT: https://www.dropbox.com/scl/fi/0v9o8xcpppdwnkntj59oi/ARCv2.pdf?rlkey=luqb6f141976vra6zdtptv5uj&dl=0 TOC: 1. ARC v2 Core Design & Objectives [00:00:00] 1.1 ARC v2 Launch and Benchmark Architecture [00:03:16] 1.2 Test-Time Optimization and AGI Assessment [00:06:24] 1.3 Human-AI Capability Analysis [00:13:02] 1.4 OpenAI o3 Initial Performance Results 2. ARC Technical Evolution [00:17:20] 2.1 ARC-v1 to ARC-v2 Design Improvements [00:21:12] 2.2 Human Validation Methodology [00:26:05] 2.3 Task Design and Gaming Prevention [00:29:11] 2.4 Intelligence Measurement Framework 3. O3 Performance & Future Challenges [00:38:50] 3.1 O3 Comprehensive Performance Analysis [00:43:40] 3.2 System Limitations and Failure Modes [00:49:30] 3.3 Program Synthesis Applications [00:53:00] 3.4 Future Development Roadmap REFS: [00:00:15] On the Measure of Intelligence, François Chollet https://arxiv.org/abs/1911.01547 [00:06:45] ARC Prize Foundation, François Chollet, Mike Knoop https://arcprize.org/ [00:12:50] OpenAI o3 model performance on ARC v1, ARC Prize Team https://arcprize.org/blog/oai-o3-pub-breakthrough [00:18:30] Chain-of-Thought Prompting Elicits Reasoning in Large Language Models, Jason Wei et al. https://arxiv.org/abs/2201.11903 [00:21:45] ARC-v2 benchmark tasks, Mike Knoop https://arcprize.org/blog/introducing-arc-agi-public-leaderboard [00:26:05] ARC Prize 2024: Technical Report, Francois Chollet et al. https://arxiv.org/html/2412.04604v2 [00:32:45] ARC Prize 2024 Technical Report, Francois Chollet, Mike Knoop, Gregory Kamradt https://arxiv.org/abs/2412.04604 [00:48:55] The Bitter Lesson, Rich Sutton http://www.incompleteideas.net/IncIdeas/BitterLesson.html [00:53:30] Decoding strategies in neural text generation, Sina Zarrieß https://www.mdpi.com/2078-2489/12/9/355/pdf

Mohamed Osman joins to discuss MindsAI's highest scoring entry to the ARC challenge 2024 and the paradigm of test-time fine-tuning. They explore how the team, now part of Tufa Labs in Zurich, achieved state-of-the-art results using a combination of pre-training techniques, a unique meta-learning strategy, and an ensemble voting mechanism. Mohamed emphasizes the importance of raw data input and flexibility of the network. SPONSOR MESSAGES: *** Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. They are hiring a Chief Engineer and ML engineers. Events in Zurich. Goto https://tufalabs.ai/ *** TRANSCRIPT + REFS: https://www.dropbox.com/scl/fi/jeavyqidsjzjgjgd7ns7h/MoFInal.pdf?rlkey=cjjmo7rgtenxrr3b46nk6yq2e&dl=0 Mohamed Osman (Tufa Labs) https://x.com/MohamedOsmanML Jack Cole (Tufa Labs) https://x.com/MindsAI_Jack How and why deep learning for ARC paper: https://github.com/MohamedOsman1998/deep-learning-for-arc/blob/main/deep_learning_for_arc.pdf TOC: 1. Abstract Reasoning Foundations [00:00:00] 1.1 Test-Time Fine-Tuning and ARC Challenge Overview [00:10:20] 1.2 Neural Networks vs Programmatic Approaches to Reasoning [00:13:23] 1.3 Code-Based Learning and Meta-Model Architecture [00:20:26] 1.4 Technical Implementation with Long T5 Model 2. ARC Solution Architectures [00:24:10] 2.1 Test-Time Tuning and Voting Methods for ARC Solutions [00:27:54] 2.2 Model Generalization and Function Generation Challenges [00:32:53] 2.3 Input Representation and VLM Limitations [00:36:21] 2.4 Architecture Innovation and Cross-Modal Integration [00:40:05] 2.5 Future of ARC Challenge and Program Synthesis Approaches 3. Advanced Systems Integration [00:43:00] 3.1 DreamCoder Evolution and LLM Integration [00:50:07] 3.2 MindsAI Team Progress and Acquisition by Tufa Labs [00:54:15] 3.3 ARC v2 Development and Performance Scaling [00:58:22] 3.4 Intelligence Benchmarks and Transformer Limitations [01:01:50] 3.5 Neural Architecture Optimization and Processing Distribution REFS: [00:01:32] Original ARC challenge paper, François Chollet https://arxiv.org/abs/1911.01547 [00:06:55] DreamCoder, Kevin Ellis et al. https://arxiv.org/abs/2006.08381 [00:12:50] Deep Learning with Python, François Chollet https://www.amazon.com/Deep-Learning-Python-Francois-Chollet/dp/1617294438 [00:13:35] Deep Learning with Python, François Chollet https://www.amazon.com/Deep-Learning-Python-Francois-Chollet/dp/1617294438 [00:13:35] Influence of pretraining data for reasoning, Laura Ruis https://arxiv.org/abs/2411.12580 [00:17:50] Latent Program Networks, Clement Bonnet https://arxiv.org/html/2411.08706v1 [00:20:50] T5, Colin Raffel et al. https://arxiv.org/abs/1910.10683 [00:30:30] Combining Induction and Transduction for Abstract Reasoning, Wen-Ding Li, Kevin Ellis et al. https://arxiv.org/abs/2411.02272 [00:34:15] Six finger problem, Chen et al. https://openaccess.thecvf.com/content/CVPR2024/papers/Chen_SpatialVLM_Endowing_Vision-Language_Models_with_Spatial_Reasoning_Capabilities_CVPR_2024_paper.pdf [00:38:15] DeepSeek-R1-Distill-Llama, DeepSeek AI https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B [00:40:10] ARC Prize 2024 Technical Report, François Chollet et al. https://arxiv.org/html/2412.04604v2 [00:45:20] LLM-Guided Compositional Program Synthesis, Wen-Ding Li and Kevin Ellis https://arxiv.

In der heutigen Folge von Künstlich Intelligent gibt es wieder spannende Neuigkeiten aus der Welt der KI: OpenAI führt eine Erinnerungsfunktion für ChatGPT ein, die Nutzern hilft, Informationen dauerhaft im Kontext zu behalten. Flux stellt euch eine Methode zum Finetunen bereit, sodass ihr zukünftig Bilder im Stile eurer Brand erstellen könnt. Das deutsche Roboter-Startup Neura Robotics sichert sich beeindruckende 120 Millionen Euro für seine nächste Wachstumsphase. Auch bei Sony gibt es spannende Entwicklungen: Ein neues Patent ermöglicht es, Spieleingaben vorherzusagen und das Gaming-Erlebnis noch smarter zu machen. Gleichzeitig sorgt Stability AI mit einem Modell zur Erstellung von 3D-Objekten für Aufsehen in der kreativen Szene. Der bekannte KI-Entwickler François Chollet startet mit Ndea ein neues Projekt, das die KI-Forschung auf den Kopf stellen könnte. Und schließlich: KI-generierte Bilder der Parteiprogramme – ein ungewöhnlicher und kreativer Blick auf die Politik. Hör rein und bleib auf dem neuesten Stand der KI-Innovationen! P.S. diese Beschreibung wurde von ChatGPT-4o erstellt. ___________ Hier noch die Links aus der Folge: Zur Neura Robotics Homepage: https://neura-robotics.com/de Artikel über die Flux Finetune API: https://the-decoder.de/eigenes-bildmodell-mit-fuenf-beispielen-flux-bekommt-eine-finetuning-api/ Direkt zum Finetune Guide von Black Forest Labs: https://docs.bfl.ml/finetuning/#implementation-guide Homepage von Devin AI: https://devin.ai/ Artikel zum neuen 3D Modell von Stability AI und Nvidia: https://the-decoder.de/neues-3d-modell-von-stability-ai-soll-schnell-genug-fuer-echtzeit-sein/ Artikel mit den KI generierten Bildern der Parteiprogramme zur Bundestagswahl 2025 von Max Mundhenke: https://www.zdf.de/nachrichten/politik/deutschland/kuenstliche-intelligenz-wahlprogramm-parteien-bundestagswahl-100.html ___________________ Social Media & Email: Hier gehts zu meinem Instagram Account: https://www.instagram.com/kuenstlichintelligent_/ Und hier zum YouTube Kanal: https://www.youtube.com/@KuenstlichIntelligent Fragen und Feedback könnt ihr gerne an meine E-Mail schicken: kontakt@torbengabriel.de

On today's episode, we discuss the latest news in the world of artificial intelligence, including: Keras creator François Chollet's new AGI lab, Microsoft expanding access to its AI assistant platform with a free tier, and Luma Labs releasing its next-generation AI video model, Ray 2. Plus, we cover Alphabet's free AI tools for Workspace, Nvidia's investment in digital copies, and a bold plan to bring woolly mammoths back with AI!

A Daily Chronicle of AI Innovations on January 16th 2025 🛠️OpenAI Launches ChatGPT ‘Tasks’ 🧪François Chollet Founds New AGI Lab 💻Microsoft Expands Copilot Access with Free Tier 🔍Contextual AI Releases State-of-the-Art RAG Platform 🧠 Researchers Develop Deep Learning Model to Predict Breast Cancer AI Fundamentals Workshop Dive into the world of AI with our comprehensive workshop. Covering basic concepts, current applications, and future trends, this session is perfect for teams looking to understand the potential of AI in their industry. SIgn up here: https://buy.stripe.com/14keV6555cAagfefYY