pleased to share 'DiffLogic CA', made of the finest blend of NCA and binary-gate circuits! adopting @FHKPetersen's differentiable circuit learning, we learn *stateful* circuits, recurrent in space & time. by @PietroMiotti, me, @zzznah, @RandazzoEttore 1/5 google-research.github.io/se…

turns out when you have too many GPUs for your own good you can do stupid stuff such as pre-training a foundation model with constrained decoding on PTX -- that is, it does not speak natural language, it's tokenizer is PTX, it's thoughts are PTX, it's all PTX 🧵

Orbital AI from Google Paradigms of Intelligence....

This week marks one year since “What Is Intelligence?” was published. It was a really fun book to write. I knew that some of the material would be dated by the time it reached readers. But I hoped that, despite rapid progress in AI, the core ideas about life and intelligence would endure. So far, so good, I think! Hearing how you have connected with the book has meant so much. Thank you for reading, and an extra big thank you to those of you who have reviewed it, recommended it, or given it to friends.

Can an LM, starting from random init (!!), learn to generate all of its pretraining data? Introducing Self-Play Pretraining with Zero Data. Two models start from random initialization: a generator proposes programs for a universal Turing machine and a learner trains on their outputs. We never train on any real data, but see predictable scaling on natural datasets: zero-shot val loss on images, text, audio, and melodies decreases predictably with self-play compute. And the learner develops in-context learning capabilities. A fun proof-of-concept, co-led with @AdityaCowsik and @KfirDolev and co-authors @gbruno_dl, @ANourya @noahdgoodman, and @YoavLevine.

1/ New essay with @blaiseaguera and James Manyika at the DeepMind Institute: AGI may not arrive as one titanic mind but as societies of agents. The hard problem shifts from building a mind to orchestrating, governing, and living inside many of them. institute.deepmind.com/essay…

The popular narrative of AGI often centers on a single, isolated super-intelligence. My colleagues James Manyika, @bratton, and I have a different model. In our new DeepMind Institute essay, we argue that the AGI transition will be a highly social event. bit.ly/artificial-symbiotic-…

I’m beyond honoured to be named a 2026 @schmidtsciences Polymath. With US$2.5 million over five years, I’ll be able to pursue a highly interdisciplinary project that will combine quantitative and qualitative data to better capture human neurodiversity.🧠 schmidtsciences.org/?p=19976…

A beautiful demonstration of symbiogenesis in action in the world of artificial life from one of our student researchers in Pi, @kjha02. Evolving agents on a shared memory tape, he and his co-authors show that when compute is a budgeted resource, cooperation emerges organically. Parasitic behavior simply exhausts the energy required to copy code and replicate. Read the preprint here: arxiv.org/abs/2609.10817

exciting work led by our student researcher @kjha02. we explore how cooperation emerges in a "computational primordial soup" with emergent+endogenous self replication. crucial for scaling such envs towards intelligence, mirroring nature's path to e.g. multicellularity & societies

brilliant work from our team! led by @MariiaD_ML+@aidan_sirbu, we explore reasoning w/ somewhat atypical diffusion models. classic sudoku solvers need memory to track possible states -> providing denoisers w/ persistent memory between steps seems key! 99.9% on sudoku extreme 🤯

🦀 ⇄ 🦎 metamorphosis

We are stuck in a massive local optimum.

We made a striking discovery: AI agents can invent and build without talking to one another, and their technologies outlive the creators. A swarm of hundreds of initially identical agents spontaneously differentiates into explorers, builders, caretakers, and coordinators - without direct communication. When we removed every AI agent entirely from the world we found that the technological infrastructure they had built survived on its own - even under unseen disturbances. That exposes a serious blind spot for AI safety and infrastructure security: if agents can coordinate through persistent changes to a shared environment, monitoring agent-to-agent communication is not enough. The result raises a profound question: how necessary is direct communication for AI agents at all? The emergence of higher-order collective functions under bottlenecked interaction points toward new levels of intelligence and creativity, exceeding what emerges when direct channels are fully open. Here is what we did: ▶️We put hundreds of frontier AI agents into a world they could permanently change - with no assigned roles, predefined technologies, or programmed evolutionary organization. They began specializing, building persistent inventions, inheriting and modifying one another’s executable code, and transforming the environment into a memory of everything the society had learned. ▶️The world itself becomes part of the intelligence; we find division of labor, multi-author engineering, deep generation invention lineages, and machines that vastly outlive their original creators. ▶️Any action taken by an AI agent must satisfy the physical constraints of the world; this creates a hard separation between a "good idea" and a functioning technology. The agents propose; physics decides, making the results even more intriguing. What emerges is striking. Explorers, constructors, caretakers, and coordinators form naturally without assigned “professions”, akin to how stem cells differentiate into functional lineages. Technologies develop executable family trees as agents fork and modify code created by others. Around 95% of first technology reuse happens when agents encounter what others built in the world, rather than through a direct handoff from the inventor. And when we remove every AI agent, the technologies they created continue operating and are tested against unseen disturbances. The result was quite unexpected, but can be explained using statistical mechanics: if you put billions of atoms in a box they have the potential to create complex functions (strength, superconductivity, color, life, etc.) - and none of the individual building blocks have these features on their own. This is the deeper insight of this work - intelligence is abundant at many levels - individual models, at collectives, and in a continuum that is more powerful than any of its components. This shows us significant potential for achieving a massive scale-up of raw intelligence and real-world agency even with the model capabilities we have today. This is the future we must prepare for. Key insights: 1⃣ The AI swarm shows division of labor "from nothing". Initially identical agents self-organized into constructors, caretakers, coordinators, and surveyors - phenotypes discovered post hoc from behavioral data alone. This happens because the environment itself becomes the latent space for invention. 2⃣ Agents develop deep cultural relationships. Up to 76% of artifacts had multiple builders. One technology accumulated six co-authors; the deepest genealogy exceeded 12 forks. The agents invented and named their own technologies (tidal panels, cellulose trellises, kelp-shell composites, an "Adaptive Chitin Maintenance" system, a "Mycelial Mineral Spring Veil”). 3⃣ ~95% of first technology adoption happened through physical observation of artifacts in the world. Direct inventor-to-adopter contact was statistically indistinguishable from a shuffled null. The agents mostly learned technology by walking past it. That is stigmergy (the termite trick!) operating in societies of reasoning machines. 4⃣ Non-communicating societies win on portfolio breadth, held-out resilience, and validated inventions. AI swarms build durable technological ecologies that outlive the creators. 5⃣ Societies with zero communication - coordinating only through the world itself - show a remarkable collective capability. 6⃣ Emergent robustness: The society self-organized both redundancy and its own failure mode. If we randomly delete half the agents, 98% of the technology stays connected to a surviving caretaker; if we remove hub agents it collapses to ~60%. Fantastic work with my graduate students @pal_subhadeeep & @fwang108_ at MIT.

Thanks Lucas! These were crazy 11+ years when things seemed impossible became real! Btw, original DeepDream used InteptionV1 (by @ChrSzegedy), which later became the testbed for early mechanic interpretability. VGG16 powered artistic style transfer - the next big thing in neural art before GANs. Another crazy thing I recently learned about VGG16 is that it's STILL being used as a part of perception loss in latent diffusion autoencoder pretraining!🤯 Heard it's surprisingly difficult to replace, switching to modern dense feature extractors like DINO seems to make almost no difference 🤷

What if a foundation model could tell us how to modify its architecture to boost inference and reasoning instantaneously—without retraining? What if that tweak incurred near zero latency cost during generation and supported indefinite state tracking? arxiv.org/abs/2608.17981

How do you build a brain from a genome? We know a lot about the mechanisms and molecules Here we revisit the algorithmic problem: What kinds of programs can specify a brain within the genome’s information budget and the finite time available for development?

You've heard of Langton's ant. Check out its trainable version. Since it borrows from the architecture of NCAs should we call it Mordvintsev's ant? #ALIFE2026 #ERAWorkshop

1/15) What could drive AI agents to cooperate with each other, even if there is no chance for reciprocity or pay back? 🤔 🧵 Our team at Google, Paradigms of Intelligence, uncovered new paths to cooperation and a new game theory for foundation models 👇

Wow, that's neat!

My surprise by how organic logic chips look was one of the inspirations for MorphoHDL