Multimodal AI
Multimodal AI breakthroughs and products
Articles
Six Wild Ideas for Perpetuating Your Mindβ10
Neuroscientists and technologists pursue cognitive immortality through six wild ideas, from connectome preservation to whole brain emulation, exploring whether ...
Computer Science > Computer Vision and Pattern Recognitionβ9
Abstract page for arXiv paper 2610.19007: Action Jets: Learning the Local Shape of the Action-conditioned Data Manifold
The Download: Synthetic Blood Breakthroughs and the Nextβ8
A $73 million DARPA program is racing to create shelf-stable synthetic blood for the battlefield, plus the next generation of batteries.
How Vision Transformers Represent Position: A Mechanisticβ9
A mechanistic study of Vision Transformers reveals how absolute and rotary positional encodings represent spatial position differentlyβexplicit features versus ...
The Download: Open Model Trade-offs and Geothermal Power's Promiseβ8
The Download: weighing open AI model risks against benefits, plus geothermal power's promise and a cyberattack on French supermarket fridges.
Open Models Have Real Benefits. Do They Outweigh the Risks?β8
Open-weight AI models offer transparency and faster innovation, but can they be recalled once released? We explore the risks, trade-offs, and governance debate.
AI Labs Are Buying Data from Dead Startups to Train Their Modelsβ10
AI labs are buying data from failed startups to train models, turning bankruptcy write-offs into valuable, domain-specific datasets that can't be scraped from t...
The Download: Lunar City Hurdles and Drone Traffic Controllersβ8
The US and China race to build lunar bases, drone traffic controllers manage autonomous flights, and an AI model submits a fake murder tip.
The Ecologist Revealing the Invisible Networks That Revive Rainforestsβ10
Tropical ecologist Daisy Dent reveals how fungal networks and seed dispersers let rainforests regenerate naturally, challenging flawed tree-planting programs.
A Camera-Native Stereo VR180 Dataset for Immersive Video Researchβ9
A firsthand stereo VR180 dataset with 1,211 camera-native RAW and fisheye samples, plus tools and annotations, benchmarking fisheye vs. half-equirectangular cod...