Knowledge catalogue
Expedia recently moved its ranking models to a state-of-the-art Keras 3 setup. Results: 30% faster training, and inference latency decreased by 70%. Read their writeup about the upgrade: https://mediu
Coding isn't yet another application domain -- it's the meta-skill required for AI to automatically develop its own training material, via symbolic world models. That's how the RSI loop actually kicks
François Chollet commented that claims the demise of Google were greatly exaggerated, cautioning against undervaluation. According to a Polymarket report, Sergey Brin is expected to take direct oversi
In the era of base LLM scaling (2022-2024), I believed the LLM line of research would reach a capability plateau (as later seen with base LLMs). In late 2024, after the o3 test-time compute demo, I ch
Outstanding cost-to-performance from DeepSeek GPT-5.6 Luna (Max) performance for a 1/4th of the cost on ARC-AGI DeepSeek V4 Flash from @deepseek_ai on ARC-AGI (Verified): - ARC-AGI-2: 61.4%, 0.04/task
This might be a good time to mention my recent keynotes titled: 'Agentic AI is Neurosymbolic AI' I would have assumed it was fairly obvious, but in case it's not: a million-line codebase (also known a
With agentic AI, workflows are increasingly CPU hungry. The share of cognition moving to the CPU keeps increasing. Scoop: AWS engineers have been told to conserve CPU compute to make sure the cloud gi
An accurate characterization of the arc of AI is that it is shaped by two trends: 1. Moving more and more logic to a neural model for tasks where training data can be densely sampled (e.g. the shift f
For a very long time most high-performing AI models were end-to-end neural models; vector input -> vector output, with only ultra-thin symbolic preprocessing and postprocessing layers (e.g. label deco
I would have assumed it was fairly obvious, but in case it's not: a million-line codebase (also known as a 'harness'), running at inference time, orchestrating thousands of calls to a neural network f
Outlook where frontier AI is headed next 18 months: The AI reasoning training + harness loop works if you can produce enough data and reasoning traces (via verifiers). Proven with code and math result
We re-tested GPT-5.6 Luna from @OpenAI on ARC-AGI (Verified) following its recent 80% price reduction: - ARC-AGI-2: 59.6%, 0.18/task - ARC-AGI-1: 90.7%, 0.07/task The new results match Luna's original
To address the limits of deep learning and avoid stalling, the field of AI started by applying patch (1), which started being demoed 9 months later in December 2024 and has now become completely ubiqu
Quick reminder of what's ok vs not ok with harnesses used for playing ARC-AGI-3: 1. Not okay: harnesses that were custom-made to solve the benchmark or that contain knowledge about the benchmark forma
One of the least explored questions in AI is how to build rich open-ended environments for learning. Benchmarks evaluate intelligence, environments shape it. iLands is interesting because it doesn't j
Really interesting concept -- instead of optimizing against static benchmarks or internally generated reward models, agents are evaluated through real economic interactions. Using an external market a
Super impressed with what Harvinder and Suman have built at @airtap_ai. They've essentially turned SMS into a headless agentic execution layer for your mobile apps. You just text it to run errands, an
Standard RL benchmarks are episodic and stationary, so they don't capture the the characteristics of real-world deployment. Morpheus is a new benchmark for continual learning that provides persistent
François Chollet observes that agentic coding systems have made dramatic progress over a six-month period, representing a significant shift in the landscape of AI-assisted development. The post reflec
GPT-5.6 Sol achieved a breakthrough by becoming the first verified frontier AI model to surpass performance on ARC-AGI-3, scoring 7.8% and setting a new state-of-the-art benchmark. ARC-AGI (Abstractio
While the writing style of LLMs is still as recognizable as ever, a new trend is that humans have started organically writing like them, too (which makes sense: of course you would end up imitating th
ARC Prize 2026: ARC-AGI-3 Milestone #1 Winners Congratulations to the three winners who open-sourced their top-scoring ARC-AGI-3 solutions: 1. @tufalabs - 1.21%, 25K 2. Reki - .867%, 7.5K 3. Md Boktia
Francois Chollet argues that reality operates according to principles that can be understood and manipulated through modeling, suggesting that comprehending the underlying structure of any system enab
Marginal cost is not a detail. It is the whole thing. Imagine going to an airline, and they tell you, 'yes, we can get you to 75% of these top destinations' -- great, but how much does it cost for eac
Reporting benchmark results as a single scalar percentage is insufficient for meaningful evaluation of model performance. Comprehensive benchmark reporting should include efficiency metrics alongside
Francois Chollet proposes a speculative future concept where 'Latent Space Archaeologists' will study the trained weights of 21st-century AI models as historical artifacts to reconstruct extinct knowl
This post likely reflects on American ideals of opportunity and possibility, discussing the cultural narrative that America enables people to pursue diverse ambitions and achieve goals regardless of b
That's a great speech that captures the essence of American exceptionalism. It is not based on innate superiority, power or wealth, but rather on the fact that the outcasts of the world could set foot
Francois Chollet argues that information and energy are the fundamental long-term constraints on technological and civilizational progress, rather than physical materials or other resources. This pers
In retrospect... Yep One very important thing to understand about the future: the economics of AI are about to change completely. We'll soon be in a world where you can turn test-time compute into com
This post explores a philosophical perspective on subjective time perception, suggesting that by engaging in more diverse activities and novel experiences daily, one can psychologically expand their l
Future jobs will prioritize adaptability and creativity over routine task execution, with emphasis on complex problem-framing skills. This reflects how automation and AI are shifting labor demands tow
Francois Chollet discusses how careful observation of ARC-AGI benchmark performance can enable early predictions about AI system capabilities and progress. The post likely highlights patterns or trend
Francois Chollet discusses how current workflows are increasingly incorporating Large Reasoning Models (LRMs) as control systems that direct symbolic programs and tools, representing a pragmatic but i
Eventually, much of AI will converge towards intuition-guided symbolic world modeling, i.e. deep learning-guided program synthesis. It is inevitable. Symbolic modeling lets a system construct a compac
François Chollet observed that all leading contenders on the ARC-AGI-3 benchmark employ a similar methodological approach, suggesting a convergence in techniques for solving abstract reasoning tasks.
ARC-AGI-3 is built different, it has dumbfounded almost all regular attempts so far because it's so much harder than anything that came before. It has no rules, it's agentic and has no explicit goals,
Cross-agent feedback loops are incredibly effective -- for a reason. Check out what @leon2mcp and team at @Bloome_im are building in this space: http://bloome.im Bloome lets you pull Claude, ChatGPT,
On the pod: 'Constrained Adaptive Rejection Sampling' with @ucsd_cse professor @lorisdanto. Hear how symbolic AI experts have navigated the LLM era and why the future of AI code generation depends on
François Chollet argues that the current wave of AI technology will not cause mass unemployment and predicts its impact on the labor market will be minimal. The post likely discusses economic and labo
This post by Keras creator François Chollet appears to reference a new development or tool intended to improve the quality of knowledge base entries or documentation. Without access to the specific co
François Chollet argues that art and science elevate humanity through the creative and discovery processes themselves, rather than through their final outputs or products. He emphasizes that the trans
A good chunk of Tokyo is also reclaimed from the bay, a process that continues to this day but has slowed down significantly (peak was in the 1960s-1970s alongside other major infrastructure buildups)
Autonomy isn't the ability to act without human supervision. It's the ability to *learn* without human bottlenecks in the process. A system that is fully dependent on human training data and RL enviro
If you wanted to do the same in SF, you could start by extending Treasure Island. Obviously the value of that new land would be considerably less than in SF proper. Perhaps 10x less. It would be econo
If your benchmark relies on a static dataset or sampling from a static distribution densely known at training time, then it is fundamentally measuring memorization/retrieval. Which might be fine if yo
This post discusses Tokyo Bay's land reclamation practices, specifically explaining why initial reclamation projects are constructed as isolated islands rather than connected to existing land—a strate
Agentic coding forces you to design clean interfaces and document them well. An agent cannot read the implicit mental model shared by your engineering team, it can only read your API contracts and doc
Francois Chollet argues that a software engineer's true value lies not in writing complex or clever code, but in their ability to protect the codebase from degradation, technical debt, and poor practi
When the cost of execution drops, the value of taste, strategy, and architectural vision skyrockets. Previously, you were spending most of your cognitive budget on the micro. Now you are free to focus
Edge cases and failure modes reveal the fundamental boundaries and constraints of complex systems more effectively than typical operations, making them valuable for understanding system behavior and l
The most complex phenomena arise from scalable recombination of very simple rules. Whether it's galaxies, chips, or neural networks, if you find the right primitive building blocks, the complexity tak
This is the strongest ARC-AGI-2 performance to date by an open-source model. GLM-5.2 from @Zai_org on ARC-AGI (Verified) - ARC-AGI-2: 22.8%, 0.25 - ARC-AGI-1: 77.0%, 0.19 Performance is comparable wit
AI in 2040 will not be built on the stack we are using today. It will be much closer to optimal. The current stack has 3-4 orders of magnitude of data inefficiency and 4-5 orders of magnitude of compu
Francois Chollet presents a framework for understanding different approaches to token usage in AI models: casual users maximize tokens for flexibility, competitive users minimize tokens for efficiency
In software, complexity is a tax you pay on every future change. You might think, 'as long as it works, I'm good, I don't care about aesthetics', but elegant code is about maintainability, not aesthet
This post likely explores the concept that maps don't merely represent territory passively, but actively influence how we perceive, organize, and interact with physical space—a reversal of the traditi
Market efficiency is a myth. Markets are made of people, who are irrational both individually and in aggregate. Maybe if human irrationality were completely random, it might average out to a rational
François Chollet announced that members of the Keras team would be attending the AI Systems DevLabs event in Sunnyvale and invited people to meet them there. This post indicates a scheduled appearance
With agentic coding, complexity compounds in a mechanical way: unnecessary code ends up in the codebase, moves to the context window, degrades the model's reasoning abilities, leads to more unnecessar