Twice a year, Benedict Evans produces a big presentation exploring macro and strategic trends in the tech industry. New in November 2025, ‘AI eats the world’.
This post includes the slides for that presentation as well as videos for Evans’ presentations on YouTube. — Read More
Recent Updates Page 8
Continuous Thought Machines
Biological brains demonstrate complex neural activity, where neural dynamics are critical to how brains process information. Most artificial neural networks ignore the complexity of individual neurons. We challenge that paradigm. By incorporating neuron-level processing and synchronization, we reintroduce neural timing as a foundational element. We present the Continuous Thought Machine (CTM), a model designed to leverage neural dynamics as its core representation. The CTM has two innovations: (1) neuron-level temporal processing, where each neuron uses unique weight parameters to process incoming histories; and (2) neural synchronization as a latent representation. The CTM aims to strike a balance between neuron abstractions and biological realism. It operates at a level of abstraction that effectively captures essential temporal dynamics while remaining computationally tractable. We demonstrate the CTM’s performance and versatility across a range of tasks, including solving 2D mazes, ImageNet-1K classification, parity computation, and more. Beyond displaying rich internal representations and offering a natural avenue for interpretation owing to its internal process, the CTM is able to perform tasks that require complex sequential reasoning. The CTM can also leverage adaptive compute, where it can stop earlier for simpler tasks, or keep computing when faced with more challenging instances. The goal of this work is to share the CTM and its associated innovations, rather than pushing for new state-of-the-art results. To that end, we believe the CTM represents a significant step toward developing more biologically plausible and powerful artificial intelligence systems. We provide an accompanying interactive online demonstration at this https URL and an extended technical report at this https URL . — Read More
Aging as a disease: The rise of longevity science
In October, the Roots of Progress Institute organized Progress Conference 2025 to connect people and ideas in the progress movement.
In this dispatch, medical historian Laura Mazer explores the conference’s longevity track, where researchers, economists, and entrepreneurs shared new ways to extend not just lifespan, but healthspan.
She finds that the frontier of medicine is shifting — from fighting disease to pursuing more life itself. — Read More
The Network is the Product: Data Network Flywheel, Compound Through Connection
The value of a data product is never contained within its boundaries. It emerges from the number, quality, and friction of its connections, and the signals from its produce. Connectivity is the architecture that turns isolated signals into coordinated intelligence. The mistake most teams make is assuming insight comes from accumulation, when in reality it comes from interaction. — Read More
OOP: the worst thing that happened to programming
In this article, we will try to understand why OOP is the worst thing that happened to programming, how it became so popular, why experienced Java (C#, C++, etc.) programmers can’t really be considered great engineers, and why code in Java cannot be considered good.
Unfortunately, programming is quite far from being a science (just like me), so many terms can be interpreted differently. — Read More
Towards interplanetary QUIC traffic
Have you ever asked yourself which protocols get used when downloading pictures from the Perseverance Mars rover to Earth? I hadn’t thought about that either, until I came across an intriguing message on the internet, back in April 2024:
I’m looking for someone knowledgeable of quic/quinn to help us out for our deep space IP project. Would be of part-time consulting. Please dm me if interested.
The message itself is quite short and somewhat jargon-y, so it took me a few readings to fully realize what the project was about:
— Working with QUIC: an internet protocol for reliable communication (i.e., what we typically use TCP for).
— Working with Quinn: the most popular Rust implementation of the QUIC protocol.
— Using QUIC to communicate between Earth and computers that are far, far away (e.g., other planets).
Business was going well on my end, and I didn’t have much time to dedicate to another consulting engagement, but… How could I say no to an interplanetary internet project? I had contributed to Quinn in the past1, so I felt well-equipped to help out and decided to actually do it. This article provides a record of the adventure so far. — Read More
The Bitter Lessons
The United States and China are often said to be in a “race” with one another with respect to artificial intelligence. In a sense this is true, but the metaphor manages to miss almost all that is interesting about US-China dynamics in emerging technology. Today I’d like to offer some brief thoughts about how I see this “race” and where it might be headed.
All metaphors are lossy approximations of reality. But “race” is an especially inapt metaphor for this context. A race is a competition with clear boundaries and a clearly defined finish line. There are no such luxuries to be found here. Beyond the rhyme, “the Space Race” made intuitive sense because the objective was clear: landing humans on the Moon.
Stating that there is an “AI race” underway invites the obvious follow-up question: the AI race to where? And no one—not you, not me, not OpenAI, not the U.S. government, and not the Chinese government—knows where we are headed. — Read More
Andrew Ng: LLMs as the Next Geopolitical Weapon & Do Margins Still Matter in AI?
Continuous Autoregressive Language Models
The efficiency of large language models (LLMs) is fundamentally limited by their sequential, token-by-token generation process. We argue that overcoming this bottleneck requires a new design axis for LLM scaling: increasing the semantic bandwidth of each generative step. To this end, we introduce Continuous Autoregressive Language Models (CALM), a paradigm shift from discrete next-token prediction to continuous next-vector prediction. CALM uses a high-fidelity autoencoder to compress a chunk of K tokens into a single continuous vector, from which the original tokens can be reconstructed with over 99.9\% accuracy. This allows us to model language as a sequence of continuous vectors instead of discrete tokens, which reduces the number of generative steps by a factor of K. The paradigm shift necessitates a new modeling toolkit; therefore, we develop a comprehensive likelihood-free framework that enables robust training, evaluation, and controllable sampling in the continuous domain. Experiments show that CALM significantly improves the performance-compute trade-off, achieving the performance of strong discrete baselines at a significantly lower computational cost. More importantly, these findings establish next-vector prediction as a powerful and scalable pathway towards ultra-efficient language models. Code: this https URL. Project: this https URL. — Read More
Introducing Nested Learning: A new ML paradigm for continual learning
We introduce Nested Learning, a new approach to machine learning that views models as a set of smaller, nested optimization problems, each with its own internal workflow, in order to mitigate or even completely avoid the issue of “catastrophic forgetting”, where learning new tasks sacrifices proficiency on old tasks.
The last decade has seen incredible progress in machine learning (ML), primarily driven by powerful neural network architectures and the algorithms used to train them. However, despite the success of large language models (LLMs), a few fundamental challenges persist, especially around continual learning, the ability for a model to actively acquire new knowledge and skills over time without forgetting old ones.
When it comes to continual learning and self-improvement, the human brain is the gold standard. It adapts through neuroplasticity — the remarkable capacity to change its structure in response to new experiences, memories, and learning. Without this ability, a person is limited to immediate context (like anterograde amnesia). We see a similar limitation in current LLMs: their knowledge is confined to either the immediate context of their input window or the static information that they learn during pre-training.
The simple approach, continually updating a model’s parameters with new data, often leads to “catastrophic forgetting” (CF), where learning new tasks sacrifices proficiency on old tasks. Researchers traditionally combat CF through architectural tweaks or better optimization rules. However, for too long, we have treated the model’s architecture (the network structure) and the optimization algorithm (the training rule) as two separate things, which prevents us from achieving a truly unified, efficient learning system.
In our paper, “Nested Learning: The Illusion of Deep Learning Architectures”, published at NeurIPS 2025, we introduce Nested Learning, which bridges this gap. Nested Learning treats a single ML model not as one continuous process, but as a system of interconnected, multi-level learning problems that are optimized simultaneously. We argue that the model’s architecture and the rules used to train it (i.e., the optimization algorithm) are fundamentally the same concepts; they are just different “levels” of optimization, each with its own internal flow of information (“context flow”) and update rate. By recognizing this inherent structure, Nested Learning provides a new, previously invisible dimension for designing more capable AI, allowing us to build learning components with deeper computational depth, which ultimately helps solve issues like catastrophic forgetting. — Read More
Rick's Cafe AI 