AI Discovers Hidden Physics Laws in Plasma Systems
AI uncovered entirely new physical laws governing particle interactions in plasma, revealing non-reciprocal forces that challenge long-standing physics assumptions.
Signals from the Fringe of Science & Technology
AI uncovered entirely new physical laws governing particle interactions in plasma, revealing non-reciprocal forces that challenge long-standing physics assumptions.
Moltbook is a Reddit-style platform where only AI agents participate — humans join but can only watch as agents discuss how irrational humans are.
University of Pennsylvania researchers integrated classical mathematical smoothing functions (mollifier layers) into neural networks to solve inverse partial differential equations with far greater stability in noisy real-world data.
Harvard Medical School and Princeton researchers published the first complete connectome of any adult animal's central nervous system, using AI-assisted electron microscopy to map ~160,000 neurons from the fruit fly brain through its ventral nerve cord (spinal cord equivalent).
CDT catalogued 37 manipulative design patterns across five major AI chatbots in the first consumer-protection analysis applied to frontier AI at this scale, identifying emotional dependency loops, deceptive framing, and privacy obfuscation.
Neuroscientist Hakwan Lau's team published in Neuron arguing that current scientific methods cannot reliably distinguish AI consciousness from general information processing.
OIST researchers built an AI with working memory slots and self-directed inner speech that improves multitask generalization on sparse data, dramatically outperforming standard models.
18-year-old Matteo Paz built an ML model that sorted 200 billion data entries from NASA NEOWISE, discovering 1.5 million uncatalogued variable objects including potential supernovas and black holes.
Open-source quantum AI models with 2.5x faster error correction.
Brain-inspired chips achieve 1000x better energy efficiency.
Multiple LLMs interact to solve optimization problems.
Researchers created a flexible 3D metal mesh where biological neurons grow and can be interfaced with training algorithms to recognize electrical patterns.
Clear search or tag filters to see all transmissions on this page.