It would be the same math/algorithms. The paper I shared uses a diffusion policy. Images are actions. Eg if you generate a spectrogram image, you can decode it immediately to speech. Speech acts can do a lot! We tend to imagine 1D embeddings for all modalities, but 2D
RESEARCH
-
Google Gemini 2.5 Pro Dominates AI Coding Benchmarks
By
–
Google vient de sortir ce qui pourrait être le meilleur modèle d’IA pour coder de toute l’histoire.
— VISION IA (@vision_ia) 10 mai 2025
Gemini 2.5 Pro I/O explose des démos réputées impossibles pour une IA.
Je l’ai testé. Résultat : bluffant. Un saut de capacité que personne ne semble encore avoir mesuré.
Vous… pic.twitter.com/ougxQCiZByGoogle vient de sortir ce qui pourrait être le meilleur modèle d’IA pour coder de toute l’histoire. Gemini 2.5 Pro I/O explose des démos réputées impossibles pour une IA. Je l’ai testé. Résultat : bluffant. Un saut de capacité que personne ne semble encore avoir mesuré. Vous
-
Shared Story Universes Shape Public Values More Than Individual Wisdom
By
–
There is a multitude of stories; the shared story universes have greater influence on public values than individually defensible wisdom
-
Can Civilization’s Systems Commit to Best Practices?
By
–
Are the nervous systems, immune response and reproductive organs of our civilization still committed to determining and following best practices?
-

Microsoft MARK: Memory-Augmented LLMs Without Retraining
By
–
Microsoft’s MARK: Memory-Augmented LLMs Microsoft introduces MARK (Memory-Augmented Refinement of Knowledge), a new framework that lets LLMs evolve with domain knowledge without retraining. Inspired by Society of Mind, MARK uses specialized memory agents: Residual
-

Jim Fan on NVIDIA’s Roadmap for Embodied AI Robots
By
–
How do we develop general-purpose robots?
— 机器之心 JIQIZHIXIN (@jiqizhixin) 10 mai 2025
A must-watch talk by Jim Fan, NVIDIA’s Director of Robotics and Distinguished Scientist, explores this pressing question.
The Physical Turing Test: Jim Fan on Nvidia's Roadmap for Embodied AI
Link: https://t.co/YGaQgl6hun pic.twitter.com/G4mfavMreEHow do we develop general-purpose robots?
A must-watch talk by Jim Fan, NVIDIA’s Director of Robotics and Distinguished Scientist, explores this pressing question. The Physical Turing Test: Jim Fan on Nvidia's Roadmap for Embodied AI
Link: https://
youtube.com/watch?v=_2NijX
qBESI
… -

Meta-Agentic AGI Demo v3 with AZR Self-Curriculum Engine
By
–
[ Meta‑Agentic α‑AGI Demo v3 — AZR‑Powered “Alpha‑Factory v1” ] Absolute Zero Reasoner (AZR) self‑curriculum — a reinforced self‑play engine that perpetually invents and solves its own tasks, unlocking open‑ended cross‑domain reasoning. GtHub: https://
github.com/MontrealAI/AGI
-Alpha-Agent-v0/tree/main/alpha_factory_v1/demos/meta_agentic_agi_v3
… -
Epigenetic Reprogramming: A Path to 10x Breakthrough?
By
–
why don’t you think epigenetic reprogramming can hit 10x? seems like an actual path to a 10x even if not in the short term.
-
Stripe’s AI Model: Excellence and Statistical Literacy Lessons
By
–
it seems simultaneously true that (A) stripe built a very good model using cutting-edge AI and (B) this is a teaching moment about basic statistical literacy
-

Model Achieves 100% Fraud Detection Rate Beyond Stripe
By
–
just coded up a model that gets even higher fraud detection rate than stripe's transformer (97->100%). AMA