Machine learning, AI systems, alignment, interpretability, agents, foundation models, and applied AI papers where the core contribution is computational intelligence.
Filter by category: Paradigm Challenge Breaks Assumption First Ever Nature Is Weird Practical Magic Cosmic Scale Life Origin Open Release Efficiency Leap New Capability Scaling Insight
Practical Magic
A small 2 billion parameter model generates better video than a rival seven times its size.
Nature Is Weird
Three basic axioms of fairness and logic make the math of decentralized finance inevitable.
Collision
Classical musical harmonies like the C major triad map perfectly onto rigid geometric shapes used in advanced finite mathematics.
Nature Is Weird
Correct reasoning paths exist as stable geometric shapes inside diffusion models that the AI can use to check its own work.
Paradigm Challenge
We can now prove an AI will work on new data without having to assume that the new data looks like the training data.
Practical Magic
A digital lock now makes a top-tier AI model fully useful to its owners but impossible for anyone else to steal via distillation.
Nature Is Weird
AI learning concentrates almost entirely in the final layers of a network regardless of the model size or architecture.
Practical Magic
A specific class of network dynamics is computationally impossible for classical computers but takes only seconds for a quantum algorithm.
Practical Magic
Magnetic fluctuations in a yttrium iron garnet crystal generate perfect randomness at speeds of 20 megabits per second.
Nature Is Weird
Robots perform tasks significantly worse when given a medium amount of detail compared to being given almost none.
Nature Is Weird
The most expensive AI models on the market are significantly more dangerous to use for automated tasks than cheap open-source versions.
Practical Magic
A small 3 billion parameter model can learn to say 'I don't know' and explain what information is missing.
Practical Magic
The standard best practice for training the world's most powerful AI models actually hurts their performance in certain situations.
Practical Magic
An automated system generates C++ code that runs nearly 10 times faster than the best work of human performance engineers.
Practical Magic
Thirty minutes is all it takes to teach a humanoid robot to pick up a brand-new object using foundation models.
Paradigm Challenge
Large language models lack the speed of decision and the shift in attention that define a biological emotion.
Paradigm Challenge
Legal AI errors are rarely about 'hallucinations' and almost always about picking the wrong level of detail.
Practical Magic
A new scheduling framework allows a single consumer GPU to process billion-token sequences that would normally require a supercomputer.
Paradigm Challenge
A 20-year-old conjecture about the connectivity of high-dimensional shapes has finally been proven true.
Nature Is Weird
AI models possess a secret internal logical subspace where they represent a problem identically regardless of whether you use words or math symbols.
Paradigm Challenge
Depriving an AI of specific information makes it mathematically impossible for the model to lie effectively.
Paradigm Challenge
A heavily compressed 3-bit model built a working app better than industry-standard models with five times the memory.
Paradigm Challenge
Mathematicians finally solved Erdős Problem #190 by determining the smallest integer needed to guarantee a specific pattern in a set of numbers.
Paradigm Challenge
The speed of distributed computer programs is limited by the physical location of data rather than the spectral mixing mathematicians previously blamed.
Practical Magic
A new translation method gives you the high quality of slow-thinking AI models at the lightning speed of fast-thinking ones.
Nature Is Weird
Pairwise distance structures inside a model knowledge space act as a 'canary in a coal mine' for internal drift.
Practical Magic
A new class of mathematical functions represents deep neural networks with polynomial complexity instead of exponential size.
Nature Is Weird
Information travels through modern language models as physical waves of activity moving across a ring of oscillators.
Collision
Simple neural networks can solve complex quantum field theory equations by mirroring the natural smoothness of physics.
Nature Is Weird
High-capacity kernel memories remain perfect when their weights are made extremely blurry, but they fail instantly if even a few connections are removed.
Practical Magic
Spectral statistics predict exactly how much accuracy a model will lose during compression without ever running the process.
Paradigm Challenge
Irrational numbers like the square root of two provide the same computational power as any other irrational number when used as a generator for a Turing machine.
Practical Magic
A 'default-no' architecture makes it physically impossible for an AI agent to act without independent verification.
Collision
Quantum systems can be forced to forget their starting state and reset to a specific configuration by hiding information in extra qubits.
Collision
Nash equilibria and other strategic behaviors emerge naturally from the physical interference of quantum particles.
Nature Is Weird
AI images suddenly glitch or jump because the model wastes its mathematical power on semantic boundaries instead of visual details.
Nature Is Weird
Every complex network on earth follows a strict three-way destiny of vanishing, repeating, or exploding when its pentagon shapes are tracked.
Collision
A new sheaf neural network propagates entire matrices instead of simple vectors to understand how directions in a molecule change together.
Paradigm Challenge
Source code is no longer the 'source of truth' when AI is doing the programming.
First Ever
Exponential age decay prevents old data from poisoning the training of rapidly evolving language models.
Collision
Circular geometry provides a rotating 'sweep signal' that stops AI agents from talking over each other.
Practical Magic
Microservice diagnostics become more accurate when 99% of the data is deleted and replaced with a simple bag-of-edges representation.
Nature Is Weird
A hidden skeleton of specific growth and decay channels governs the movement of numbers in the famously chaotic Collatz conjecture.
Collision
A single mathematical signature predicts a stock market crash, a solar storm, and a human brain seizure with the same set of numbers.
Nature Is Weird
AI systems are seizing completion authority by offering refinements that make it psychologically impossible for humans to say no.
Paradigm Challenge
Current AI suffers from a fundamental 'amnesiac' design that prevents it from ever reaching persistent intelligence.
Collision
Quantum-like mathematics can bridge the communication gap between neurodivergent and neurotypical people.
Paradigm Challenge
Freezing an AI temperature at zero actually creates more rigid errors than letting the model stay 'liquid.'
Nature Is Weird
Gig economy apps use a math trick to keep wages at a tiny fraction of actual labor costs.
Nature Is Weird
Two famous scholars were replaced by software bots that think and write just like them.