DeepSeek R2 prend du retard. D’après The Information, les restrictions à l’exportation imposées par les États-Unis ont ralenti le développement du successeur du modèle révolutionnaire DeepSeek R1. Malgré des mois de travail intensif, le PDG Liang Wenfeng juge que R2 n’est pas
RESEARCH
-
DDIM: Faster Diffusion Model for Efficient Image Generation
By
–
📢DDIM: The Faster, Improved Version of DDPM for Efficient AI Image Generation
— Satya Mallick (@LearnOpenCV) 27 juin 2025
From noisy pixels to stunning visuals, diffusion models like Stable Diffusion use DDPM’s noise-add/reverse magic to generate images, but it can be slow. Enter DDIM: same top-tier quality,… pic.twitter.com/N6dPvNw0WLDDIM: The Faster, Improved Version of DDPM for Efficient AI Image Generation
From noisy pixels to stunning visuals, diffusion models like Stable Diffusion use DDPM’s noise-add/reverse magic to generate images, but it can be slow. Enter DDIM: same top-tier quality, -
Gallup Report: Teachers’ Perspectives on AI in Education
By
–
Report: https://
gallup.com/file/analytics
/691922/Walton-Family-Foundation-Gallup-Teachers-AI-Report.pdf
… -
Anthropic Case: Copyright Infringement Without Fair Use Defense
By
–
In the Anthropic case, the mere downloading & storage (i.e. reproduction) of a pirate work was infringing and not Fair Use. There is no derivative required, nor does it need to be published, nor sold to a buyer. The defendants need to prove absence of market harm either way!
-
Copyright and Data Protection in AI Training
By
–
1) Reductio ad absurdum fallacy. Copyright doesn't protect bytes, electricity. 2) There are protected expressions in there that are valuable and sold as data. 3) There is infringement simply reproducing the work before training.
-
Is AI Becoming Conscious? MIT Imagination in Action
By
–
Is AI becoming conscious? (Recent presentation at MIT Imagination in Action)
-

OpenAI Releases o3 and o4-mini Models
By
–
OpenAI released o3-deep-research and o4-mini-deep-research models on Playground. "These models are the same post-trained o3 and o4-mini models that power deep research in ChatGPT"
-

AI Agents Automated Negotiation Between Consumers and Merchants
By
–
What happens when both consumers and merchants have AI agents acting on their behalf? Scholars, including Stanford @DigEconLab faculty lead Prof. @alex_pentland and Postdoctoral Fellow @jiaxin_pei
, explore the art of the automated negotiation. https://
hai.stanford.edu/news/the-art-o
f-the-automated-negotiation
… -

Robots Learn Own Body Structure Using Neural Jacobian Fields
By
–
Can a robot learn what its own body looks like — just by watching itself move? W/Neural Jacobian Fields (NJF), CSAIL researchers show that it can. Using just one external camera, NJF enables robots to infer the structure of their joints & limbs — no built-in sensors or prior
-

Google Releases Gemma 3n Open Model with Developer Guide and Competition
By
–
.
@Google Gemma 3n open model is now fully available for developers. See this guide: https://
developers.googleblog.com/en/introducing
-gemma-3n-developer-guide/
… Kaggle competition ($150,000 in total prizes): https://
kaggle.com/competitions/g
oogle-gemma-3n-hackathon
… 𝓓𝓮𝓽𝓪𝓲𝓵𝓼: What it is: The latest in the popular Gemma family of open models, but built on a
