Amaury de Buchet

Social media is populist and polarising; AI may be the opposite ft.trib.al/YuTNwyS | opinion

Hmmmmmmmmmm. ft.com/content/1171d6…

AI might be a powerful depolarizer in political discourse

Every LLM from any lab today traces back to this guy, who was the only person at OpenAI pushing for pretraining transformer language models. He built GPT-1. After that did others see the potential. He invented it, and almost none of the so called AI experts even know his name.

🎮 Saviez-vous que Paris IDF est la 1ère région française du jeu vidéo avec 300 sociétés spécialisées, soit 42 % des acteurs nationaux et + de 4 000 emplois? ▶️Découvrez l’étude du Crocis « L’IDF, un territoire attractif pour les entreprises du jeu vidéo » cci-paris-idf.fr/fr/prospective…

Sam Altman veut nous vendre de l’intelligence comme on nous vend de l’électricité… et l’histoire lui donne raison ! En 1882, Thomas Edison n’a pas seulement inventé l’ampoule. Il a construit un réseau électrique, installé des compteurs, puis facturé son usage mensuellement. Rockefeller a fait pareil avec le pétrole, AT&T avec les appels téléphoniques et AWS avec le cloud… Le secret c’est pas l’invention. C’est le réseau… et le compteur qui va avec. Ce qu’on va payer demain c’est la capacité de rivaliser avec les autres. Celui qui peut payer pourra lancer une startup, créer plus vite, apprendre plus vite, décider plus vite. Celui qui ne peut pas… continuera à chercher sur Google. La Silicon Valley vous vend de la valeur ajoutée sous forme d’abonnement mensuel. Et bien sûr que vous allez payer. Parce qu’à défaut vous risquez d’être le plus con dans la pièce.

Anthropic didn't reject the Pentagon. They rejected the blank check for dystopia. Dario drew two lines: no mass domestic surveillance, no fully autonomous kill decisions. Everyone else whispered yes for the money. Anthropic shouted no in public, linked the blog, and dared the DoD to walk. This isn't purity theater. It's the first lab proving safety isn't just internal red-teaming—it's refusing the state's kill-chain upgrade when the ask crosses into civilian panopticon or Skynet territory. Pentagon thought pressure would fold them. Deadline came. Anthropic held. Now the choice is theirs: blacklist Claude from classified nets and look petty, or swallow the limits and admit even the war machine has boundaries when a company has spine. The old playbook—pay, pressure, procure—hit a wall. Most labs would have bent. Anthropic bent the narrative instead. Your frontier model isn't just compute anymore. Some still treat it like a moral lever. The rest are already rewriting terms of service for the highest bidder.

🚨BREAKING: Google just dropped another hit! It's called PaperBanana and it generates publication-ready academic illustrations from just your methodology text. No Figma. No manual design. No illustration skills needed. Here's how it works: A team of AI agents runs behind the scenes → One finds good diagram examples → One plans the structure → One styles the layout → One generates the image → One critiques and improves it Here's the wildest part: Random reference examples work nearly as well as perfectly matched ones. What matters is showing the model what good diagrams look like, not finding the topically perfect reference. In blind evaluations, humans preferred PaperBanana outputs 75% of the time. This is the recursion we've been waiting for AI systems that can fully document themselves visually. Waitlist’s open, Link in the first comment.

Can you read 900 words per minute? Try it.

BREAKING: I stopped watching 2-hour YouTube tutorials. Grok now turns any video into a step-by-step skill plan. Here are 8 prompts that convert videos into mastery:

🚨 RAG is broken and nobody's talking about it. Stanford just exposed the fatal flaw killing every "AI that reads your docs" product. It's called "Semantic Collapse", and it happens the moment your knowledge base hits critical mass. Here's the brutal math (and why your RAG system is already dying):

🚀 #Digital2026 est là ! La nouvelle édition de notre série Digital Reports vous livre toutes les dernières datas et tendances mondiales : réseaux sociaux, internet, mobile, médias, IA, e-commerce… Quelques enseignements ici et en thread 👇 wearesocial.com/fr/blog/2025/1…

The good old days of critical thinking 🤣

If you're a teacher, parent, or educator, here's a quick way to turn your lessons into printable crosswords using @NotebookLM + CrosswordLabs. 1️⃣ Add your lesson materials into NotebookLM. It pulls from your sources so the content stays on point. 2️⃣ Use the prompt shown in the video, in NotebookLM( You can adjust the number of clues) 3️⃣ Take the output and paste it into crosswordlabs. It’ll auto-build your crossword. 4️⃣ Once ready, you can either print, download as PDF or share the link if you want students to solve it online.

Smallest Dell server ever 🇯🇵 minimachines.net/actu/microserv…

Voici à quoi ressemblerait l’Antarctique sans glace : cette carte révèle un paysage insoupçonné ift.tt/dSbBEGf

Introducing the Responses API: the new primitive of the OpenAI API. It is the culmination of 2 years of learnings designing the OpenAI API, and the foundation of our next chapter of building agents. 🧵Here’s the story of how we designed it: