Multimodal AI
Multimodal AI breakthroughs and products
Articles
Emergent Object Binding Has a Finite Spatial Horizon⭐9
Frozen Vision Transformer embeddings encode same-object binding only within a finite spatial horizon, decaying exponentially with patch distance, reshaping how ...
Redefining Enterprise Intelligence with Autonomous AI⭐10
Autonomous AI can unify fragmented enterprise intelligence through composable infrastructure, sovereign data, and cross-functional coordination, turning isolate...
The Download: Inside the Biological De-Aging Contest and Why⭐8
A new biological de-aging contest pits 500 competitors against each other to reverse their biological age, plus why LLMs still can't reason.
Don't Be Fooled—LLMs Don't Reason⭐10
LLMs like GPT predict the next token—they don't reason. Explore why today's AI lacks the deliberate, search-based inference that made AlphaGo possible.
The Race for Biological Youth: Inside a New Longevity Contest⭐8
A new longevity contest challenges competitors to reduce their biological age. I'm joining to document what works, what doesn't, and the science behind it.
STATERA: Hidden Mass Estimation via Zero-Shot Sim-to-Real⭐9
STATERA adapts frozen V-JEPA video features with a temporal tubelet mixer to estimate hidden center-of-mass from monocular video, improving zero-shot sim-to-rea...
The Download: AI "Mind-Reading" Breakthroughs and the Rise of⭐9
A new AI tool reconstructs what you're viewing from brain scans, raising privacy concerns, plus how distributed batteries could help stabilize the grid.
AI "Mind-Reading" Tool Reconstructs What You're Looking At From⭐9
A new AI system reconstructs images a person is viewing from fMRI brain scans alone, raising hopes of someday decoding mental imagery—and even dreams.
How Smaller, Distributed Batteries Could Help the Grid⭐9
Distributed batteries in e-bikes, stoves, and appliances are helping stabilize the grid, cut peak demand, and speed the clean energy transition.
Multi-Task Morphological Concept Learning for Pulmonary Nodule⭐8
A study evaluates multi-task 3D CNN learning of radiologist-annotated morphological features for pulmonary nodule malignancy assessment on LIDC-IDRI CT scans.