Signals from the Fringe of Science & Technology
University College London researchers combined quantum computing with AI to predict chaotic physical systems with 20% greater accuracy and 100× less memory than classical models.
Northwestern University scientists printed artificial neurons using molybdenum disulfide and graphene that generate electrical signals matching biological neurons, successfully triggering real neurons in mouse brain tissue.
Meta, UT Austin, UC Berkeley, UCL, and Harvard demonstrate that reinforcement learning follows predictable scaling laws with over 400,000 GPU-hours of empirical validation.
Researchers David O'Ryan and Pablo Gómez developed AnomalyMatch, an AI tool that scanned 100 million Hubble images in 2.5 days to find 1,400 anomalous objects—including 800+ never-documented galaxies: jellyfish galaxies, gravitational lenses, merging systems, and massive star clusters.
Google's Gemini model with extended reasoning solved expert-level unsolved problems: a decade-old optimization conjecture, network deadlock puzzles, physics singularities in cosmic strings, and extensions to auction theory—working across mathematics, physics, and economics.
Brad Aimone and Brad Theilman at Sandia National Laboratories developed algorithms enabling brain-inspired neuromorphic chips to solve sparse finite-element physics problems (fluid dynamics, electromagnetic fields, structural mechanics) with unprecedented energy efficiency—previously requiring supercomputers.
Matteo Paz, a Pasadena high school student, used machine learning on NASA's NEOWISE telescope archive to discover 1.5 million previously undetected variable light sources in infrared data.
Machine learning system identifies 100+ new planets including worlds in Neptunian desert
Mollifier Layers enable accurate high-order derivatives in neural networks without automatic differentiation
Atomically thin materials enable valley-degree-of-freedom computing for faster AI/quantum chips
Ultra-fast photonic chip harnesses light-matter particles for AI and quantum acceleration
NASA is testing a next-generation space computer chip that enables spacecraft to operate far more independently in deep space, reducing reliance on Earth-based command signals.
Clear search or tag filters to see all transmissions on this page.