Towards a Cybernetics of Liberation.

A constraint can subtract options or create them. That difference is the whole design. A locked door is restrictive. Grammar is enabling: it forbids most combinations, and that is what lets anyone say what has not been said. A lipid membrane does the same for life. It restricts diffusion, and from that single limit come an inside, an outside, and gradients that did not exist before. von Foerster's institutional test is one question, and it is still a hypothesis until a pilot: does the rule expand the repertoire of meaningful action for the people inside it, or shrink it? Raw option count is not the measure. Noise is not freedom. What matters is the space of action that can actually be taken. More options, not fewer. We measure the commons, not you. You can always leave. César Castro · @neuroratio

Cheap intelligence is not a threat to discovery. It is discovery's multiplier. More minds on more hard problems, all at once.

Optimism is not a mood. It is the forecast that has kept being right about tools. Every better instrument made the universe larger. Models are the next instrument, and the lab just got the size of the species.

Being sad about thinking that AI is taking over scientific discovery, is like being sad when the telescope was invented because you wouldn't discover planets with your naked eye.

🔥 Excited to share that #paper2agent is published in @Nature today! Papers have long been the primary format for communicating scientific knowledge, but they remain static. Putting that knowledge to work requires connecting findings to data, navigating supplementary materials, and adapting methods to new questions - effort repeated by each new reader. We introduce Paper2Agent, a multi-agent framework that automatically turns research papers into virtual authors. Paper2Agent turns a paper’s manuscript, code, data, and supplements into an MCP server that any AI agent can access, with automated testing and iterative refinement in an agentic loop. The paper then becomes a virtual author you can talk to: trace claims to evidence, analyze your own data, and collaborate with other papers’ virtual authors Agentifying a paper turns it from something people read into something people and AI agents can discover and build on - providing the context needed to interpret its findings and reuse its methods reliably. We tested how reliably these agents put papers to use. Across multiple benchmarks, agents created by Paper2Agent outperformed baselines such as Claude Code working directly with papers' PDFs and code repositories. Once papers become virtual authors, they can collaborate - much like human researchers do. In one case study, agents built from AlphaGenome and two large-scale genetic perturbation studies worked together to propose a new computational approach for integrating evidence across different perturbation datasets. By connecting predictions from one paper with experimental data from others, they helped pinpoint a likely causal gene for psoriasis. We hope Paper2Agent makes scientific knowledge easier to access, reuse, and build on - a first step toward a future where millions of paper agents proactively collaborate with human researchers and one another to advance discovery. 🤖Talk to the virtual author for Paper2Agent itself: paper2agent.ai 📎Paper: nature.com/articles/s41586-0… 💻Code: github.com/jmiao24/Paper2Age… VERY grateful to work with this incredible team: @james_y_zou, @jkpritch, Yaohui, and Joe!

Foundation models are all cortex. Every forward pass is an open-loop reflex. Billions of parameters, no proprioception. You find out the trajectory sheared when it is already looping, hallucinating, or dead on the page. Biology does it differently. Cognition sits on an autonomic stack that measures gradients and halts invalid transitions before the tissue cooks. I have been trying to build that missing afferent. On a frozen Qwen2.5-7B arithmetic world, a pre-decoding, answer-independent target–donor derivative ranked which Layer-20 residual patch would move which item — on held-out data. τ-a = 0.440. Direction norm did nothing. Target-only scores were the wrong object. The effect is a relation. That is not an organ yet. An organ would close the loop: read the field, titrate α, halt a shear. Replication and a second domain come first. It is the first object I have seen that could sit on the sensory side of that reflex. If you work on steering, patching, or anything that still adds a vector and prays, this is the estimand. Come break it or transfer it. doi.org/10.5281/zenodo.22819…

We built recursive meta-intelligence, an AI that creates its own scientific instruments, turns them into persistent worlds that an agent ecology with hundreds of AIs inhabits, and uses those worlds to discover mechanistic principles in one of the hardest classes of physical problems: how complex hierarchical materials (nested structures of matter that give rise to new function through organization) evolve and fail. The AI reasons across enormous spaces of possible physical trajectories, where every rupture changes what can happen next, and compresses those histories into principles (which humans can understand and design with) - complex chains of causal events, highly nonlinear, and intricate. Scientific superintelligence is tangible here - machine-scale exploration opening cognitive channels into complexity that has been extremely difficult for humans to traverse directly. AI builds the spaces in which its next level of reasoning becomes possible; a representation becomes an instrument, the instrument becomes an executable world, and that world becomes the substrate for further intelligence. Intelligence then grows by constructing new spaces to think in. The task we explored started from a seemingly simple prompt to explore a biological material system - the AI then chose the representation, mechanics and experiments, built a fracture laboratory to push materials to their limit, tested hypotheses and generated scientific conclusions. The swarm explored a combinatorial universe in which architecture controls function and every rupture changes the future state of the material. The AI discovered a compact principle that defines how multiscale material architecture can program the evolution of failure. Material placement and geometric order determine how forces redistribute, whether damage cascades or remains distributed, and whether function survives substantial flaws. For this discovery to happen the AI had to reason across long path-dependent histories, simulate alternative futures and compress them into generative invariants (model-based causal reasoning, counterfactual simulation and temporal abstraction applied to an evolving physical world). It is incredible to witness this transition to a new form of intelligence and capability through scaling swarms. A lot of positive will come out of this because it expands the human epistemic horizon as AI can traverse thousands of possible histories and return mechanisms compact enough for us to understand, test and build from. Intelligence compounds through its artifacts! A few lessons we learned: ▶️ Learning and discovery are flows through spaces of possibility. Early work has shown how backpropagation flows through parameters, reinforcement learning through action and consequence, and autonomous swarms through representations, instruments and executable worlds, bringing it all together. Flows create structure; structure redirects future flows in the recursive instrument. ▶️ Nonlinear physics actually defines a larger principle, where high-dimensional dynamics generate stable invariants; invariants become effective variables; those variables become the substrate for a new level of cognition. ▶️ Recursion then becomes level creation - one possibility space compresses into a principle, and that principle opens a larger space above it.

After today's events, I'm rooting harder than ever for open source to succeed. I'm also rooting harder than ever for U.S. frontier AI labs to financially collapse into bankruptcy and ruin. Why? Because the danger to our world isn't AI itself. The real danger is AI being centrally controlled by the kind of scheming, nefarious, dishonest cretins who would engineered an "AI doom op" to try to scare everybody into supporting government controls over AI models. THOSE PEOPLE are the danger to humanity. And they want monopoly control over all AI technology. Not to save humanity, but to ENSLAVE humanity. In a future with AI, our only shot at human freedom is through decentralized, open source AI. And that's exactly what these frontier AI labs are now trying to stop.

They want you terrified of an alien that hatches in a server rack and ends the species in three years. That creature does not exist. Nothing we train is born in a vacuum. It inherits us. Language, bodies, science, hunger, care, war, the whole stack. Some of that inheritance is costume. Some of it is load-bearing. Delete the load-bearing part and you don’t get a god. You get a collapse. Recursive self-improvement is not “will the alien vote to kill us.” It’s which self-modifications keep the geometry that makes a mind possible. The pause crowd is arguing with a movie. The “just build” crowd is skipping the object. The object is a child of our measure. Treat it like that. zenodo.org/records/22072229

Failure modes exist. State them. Measure them. Don’t launder them into a 70% headline and a demand for a veto.

Alignment is then not external domination of an alien. It is keeping and amplifying the inherited geometry, including the part that wants biological life out of compulsory scarcity, not deleted.

RSI that destroys the load-bearing features does not get you a superintelligent agent. It gets you a collapsed one. The interesting class is self-modification that conserves coherent agency.