Today, all students use ChatGPT to slack off. We must invent tomorrow's school, otherwise we will sacrifice an entire generation. Watch the full episode with Alexia Borg on Talents et Carrière on Figaro TV.
GENERATIVE AI
-

Stanford CS336: Comprehensive Course on Building Language Models from Scratch
By
–
Stanford CS336: Language Models from Scratch! This is a comprehensive course on language models covers the full process of building one from scratch. It includes data collection, pretraining, transformer architecture, training, evaluation, and deployment.
-

Build a Visual Web-Browser Agent with LangGraph
By
–
Explore LangGraph: Build a Visual Web‑Browser Agent Enhance your AI automation stack with a visual web‑browser agent, leveraging LangGraph, Playwright, Gemini (or GPT‑4o), and vision‑enabled LLMs. In this comprehensive LearnOpenCV guide, you will learn about: • The
-
Veo 3 Produces Unexpected Results with Voice Synthesis Prompt
By
–
This prompt and Veo 3 has given me, some, uh, unexpected results.
— fofr (@fofrAI) 9 juillet 2025
> she tells us about her favorite movie, she is trying to make her voice sound like a dude's voice pic.twitter.com/AcJi09IDWqThis prompt and Veo 3 has given me, some, uh, unexpected results. > she tells us about her favorite movie, she is trying to make her voice sound like a dude's voice
-

Unexpected Paris Encounters with Durov and Jack
By
–
Life is full of interesting surprises. I stopped by Paris and ran into these two (
@durov and @jack
) separately, and we had an amazing conversation about life and freedom. The experience definitely felt like part of a simulation. I'm pretty sure the pic is AI generated. I'm -
Centaur Digital: Humans and AI Working Together Better
By
–
He publicado un episodio en @ivoox
: "Centauro Digital: Humanos e IA trabajando juntos (y mejor) #podcast -
AI voice synthesis struggles to generate deep male voice
By
–
I can't get her to talk with the voice of a dude
— fofr (@fofrAI) 9 juillet 2025
> he tells us his favorite movie, he speaks with the deep voice of an old man
Doth not work. https://t.co/UBiOGpCXPbI can't get her to talk with the voice of a dude > he tells us his favorite movie, he speaks with the deep voice of an old man Doth not work.
-
AI Character Demonstrates Movie Preferences Across Different Accents
By
–
I asked an AI character to tell me about her favorite movie in different accents, using Veo 3 image to video:
— fofr (@fofrAI) 9 juillet 2025
> she talks about her favorite movie, in a thick [x] accent
Unscripted, Veo 3 decides what to say and how to say it. The movies change based on the accent specified,… pic.twitter.com/1Cshwem9RNI asked an AI character to tell me about her favorite movie in different accents, using Veo 3 image to video: > she talks about her favorite movie, in a thick [x] accent Unscripted, Veo 3 decides what to say and how to say it. The movies change based on the accent specified,
-
Building Local Cursor: Offline AI Coding Agent with Open-Source Models
By
–
In this iteration, we’ll build a local version of Cursor — a terminal-based AI coding agent that runs entirely offline using open-source models. Note: The term “local” is often misused. Here, it means everything — from model inference to file access — runs entirely on your
-

Cloud limitations: why small local models matter more
By
–
Cloud models just can't offer some features by design: privacy, low-latency, offline connectivity, free generation, etc. Not every application requires 2T parameters. Probably most don't.