We call for a prohibition on the development of supercolliders, not to be lifted before there is:
1. broad scientific consensus that it won't create a black hole that will swallow the universe, and
2. strong public buy-in.
RESEARCH
-
Supercollider Development Moratorium Calls for Safety Consensus
By
–
-
Jensen Huang: AI Will Let Us Converse With Living Cells
By
–
« Tu pourrais parler à une cellule comme tu parles à un chatbot. »
— VISION IA (@vision_ia) 24 octobre 2025
Jensen Huang affirme que la même IA qui comprend le langage comprendra bientôt la vie elle-même transformant la biologie en quelque chose avec laquelle nous pourrons converser.
« Quelles sont tes propriétés ? À… pic.twitter.com/DGXxlT6PEZ« You could talk to a cell the way you talk to a chatbot. » Jensen Huang asserts that the same AI that understands language will soon understand life itself, transforming biology into something we can converse with. « What are your properties? What can you bind to? »
-
Transformers and Self-Organization in Evolving AI Systems
By
–
The transformer died for our sins. The principle of self organization in an evolving universe is the alpha and the omega
-
Visual System as Generative Model: Sensory Constraints and Hallucinations
By
–
The visual system is a generative model that gets punished when it contradicts sensory patterns. In the absence of sensory patterns, it generates freely: during dreams and out of body experiences, it invents hallucinations.
-

David Cope’s EMI Algorithm Passed Music Turing Test 1990s
By
–
Except the Turing test for music was passed in the 90s by David Cope's EMI algorithm: https://
nytimes.com/2025/06/02/tec
hnology/david-cope-dead-ai-music.html
… -
Stanford Researchers Identify Error Propagation in AI Agents
By
–
Stanford researchers just solved why AI agents keep failing.
— Shubham Saboo (@Saboo_Shubham_) 24 octobre 2025
They watched 500+ agent failures across three benchmarks.
Found a pattern nobody expected:
Early mistakes don't just cause problems – they cascade into complete system meltdowns.
It's called error propagation.
One… pic.twitter.com/mQvuC3GKUjStanford researchers just solved why AI agents keep failing. They watched 500+ agent failures across three benchmarks. Found a pattern nobody expected:
Early mistakes don't just cause problems – they cascade into complete system meltdowns. It's called error propagation. One -
The Synthetic Data Trap: AI Strategy and Average Performance
By
–
The Synthetic Data Trap: How Your AI Strategy Is Training You To Be Average
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @demishassabis @Ronald_vanLoon @TamaraMcCleary @geoffreyhinton @goodfellow_ian @jeffdean @erikbryn http://
ow.ly/4Ym930sQ3r5 -

MIT’s Breakthrough in AI Safety
By
–
MIT just cracked AI safety. not with more filters. not with more rules. with one insight everyone missed. they taught models to think backwards first. enumerate every possible harm. analyze every consequence. only then respond. they call it InvThink, and it might redefine how
-

SnorkelSpatial Benchmark Tests LLM Spatial Reasoning Abilities
By
–
New benchmark drop SnorkelSpatial tests how well LLMs can think in space, following text-based moves and rotations in a 2D world.