Signals from the Fringe of Science & Technology
Researchers demonstrated that combining traditional neural networks with symbolic reasoning — mimicking how humans decompose problems into steps and categories — can cut AI energy consumption by up to 100x while actually improving model accuracy.
University of Pennsylvania physicists demonstrated all-optical switching in AI hardware using exciton-polaritons (hybrid light-matter particles) consuming only 4 femtojoules — eliminating the energy-burning electron-to-light conversion bottleneck in photonic AI chips.
A multi-institution study pitting generative AI against 100,000+ humans found AI now outperforms the average person on standardized divergent creativity tests — though top-10% human creatives still win.
Harvard SEAS researchers published in PNAS that robot swarms reach destinations faster when given a small degree of random movement — too little causes deadlock, too much causes chaos, but the sweet spot produces fluid collective flow.
A new brain-like chip using spiking neural networks can now solve complex partial differential equations — once thought limited to energy-hungry supercomputers — while cutting energy consumption by up to 70%.
MIT Technology Review named mechanistic interpretability — the field that reverse-engineers exactly which computations inside a neural network produce a given output — its #1 breakthrough technology of 2026.
Scientists successfully tested an AI-designed universal coronavirus vaccine in humans for the first time, finding it safe and well tolerated — an AI-to-bench-to-human pipeline milestone.
Cobalt — a metal physicists thought they fully understood — turns out to harbor a dense network of topological quantum states that remain stable at room temperature, potentially enabling room-temperature quantum devices.
UC San Diego researchers combined generative AI methods with physics-based data to produce climate models that run ~25x faster than conventional approaches, opening the door to real-time regional climate forecasting.
An AI system called RAVEN found more than 100 previously hidden exoplanets in NASA telescope data, including rare extreme worlds that human analysts had missed.
USC engineers built a tungsten-graphene memristor that stores data reliably at 700°C — hotter than molten lava — for over 50 hours and one billion switching cycles, published in Science.
Tsinghua University demonstrated neuromorphic computing using perovskite microcavity exciton-polaritons — hybrid light-matter quasiparticles — achieving 92% digit recognition accuracy with single-step training at room temperature.
Clear search or tag filters to see all transmissions on this page.