Françoise Morvan

Via @visibrainFR #CES2025 - Zoom sur les influenceurs et journalistes francophones les plus engageants sur X 🔎

【盲点】ロボットの手が小さな部品をそっとつまみ上げた。Shadow Robotの義手型ハンドは24の自由度と40本の人工腱で人間そっくりの指の動きを再現する。100個以上のセンサーが繊細な力加減を検知する仕組みだ。

Andrew Ng just dropped a 3-hour course on how to become an AI Engineer in 2026: 00:00 - build agentic AI systems from scratch 04:25 - where AI engineering is actually headed 23:38 - the full AI prompting course 2:52:17 - build a working app with AI in 30 minutes This 3-hour watch replaces any $1000 course you could pay for. Watch it today, then go deeper with the full guide on building AI agents below.

【発明】重い荷物を担いで跳びはねる二足歩行ロボットが発明された。Boston DynamicsのHandleは車輪の付いた脚で走り最大45キロの箱を持ち上げる。垂直に1.2メートル跳躍し時速15キロで倉庫を駆け回る。

Rien qu'au bruit du moteur, Mélanie, 25 ans, peut reconnaître n'importe quel bus à Toulouse. Depuis un an, elle partage sa passion sur les réseaux sociaux. Elle nous a donné rdv dans l'un de ses bus préférés.

Algunas personas escuchan las palabras en sus cabezas mientras leen. La lectura en silencio activa áreas auditivas del cerebro que "escuchan" nuestra voz interna, como la corteza auditiva y áreas temporales. Este estudio con pacientes epilépticos muestra que leer no solo es visual, sino una experiencia multisensorial compleja donde se integran respuestas neuronales visuales y auditivas m.youtube.com/watch?v=gZuLG0… jneurosci.org/content/32/49/…

Masasına tam 1.000.000 dolar nakit ödül koydular. Dünyanın en prestijli üniversiteleri, Harvard ve Stanford, ona açık çek sunmak için kapısında yatıyordu. Ama o, parayı ve şöhreti elinin tersiyle itti. Annesinin rutubetli, döküntü evine geri döndü ve ormana mantar toplamaya gitti. Karşınızda tarihin en gizemli ve en tavizsiz dehası: Grigori Perelman. Yıl 2002. Perelman, matematikte 100 yıldır dünyanın en parlak beyinlerinin uğraşıp da çözemediği o meşhur "Poincaré Sanısı"nı çözdü. Hem de basın toplantısı yapmadan, hiçbir dergiye kapak olmadan, sessizce internete 3 sayfalık bir PDF yükleyerek. Bilim dünyası şoka girdi. Herkes bu darmadağınık saçlı, eski püskü giyinen adamın peşine düştü. Ona "Fields Madalyası" (Matematiğin Nobel'i) vermek istediler. İspanya kralı onu sarayına davet etti. Perelman'ın cevabı buz gibiydi: "İstemiyorum." Herkes onun delirdiğini düşündü. Bir insan neden milyonlarca doları ve sonsuz şöhreti reddederdi ki? Çünkü Perelman için mesele hiçbir zaman para, statü veya insanların alkışı olmadı. O, sadece evrenin nasıl işlediğini anlamak istiyordu. Problemi çözdüğü an, oyun onun için bitmişti. Sonrasındaki ödül törenleri, kameralar ve sahte tebrikler ona sadece bir sirk gibi geliyordu. Perelman, modern dünyanın en büyük hastalığına tek başına kafa tuttu: "Onaylanma ihtiyacı". Bugün hepimiz; birileri bizi beğensin, takdir etsin, sosyal medyada alkışlasın diye kendi hayatımızı köleleştiriyoruz. İstemediğimiz hayatları, başkalarına gösteriş yapmak için yaşıyoruz. Perelman ise bize şunu gösterdi: Eğer yaptığın işin değerini gerçekten biliyorsan, başkalarının sana vereceği hiçbir madalyaya ihtiyacın yoktur. Amacı olan bir zihni, hiçbir para satın alamaz.

Lejos de gastar de inmediato los ingresos obtenidos por el petróleo y el gas, el gobierno noruego optó por invertir gran parte de esos recursos en un fondo soberano que hoy es considerado el más grande del mundo. Su objetivo es que esa riqueza beneficie no solo a la generación actual, sino también a las futuras. Aunque suele decirse que “cada ciudadano es copropietario”, en realidad el fondo pertenece al Estado y se administra en nombre de toda la población. Los rendimientos ayudan a financiar servicios públicos y a mantener una economía más estable a largo plazo. Un modelo de administración que ha convertido a Noruega en un referente mundial en el manejo responsable de sus recursos naturales.

Crazy… Air humanoid Jordan Magic Lab has released a demo: The full-sized humanoid robot MagicBot X1 performs a flying slam dunk🤾‍♂️ Another kind of aesthetic of violence…

This youtu.be/atcqMWqB3hw?si…

There is one piece of software running on 20 billion devices right now. Your phone uses it. Your car uses it. Your TV uses it. Your router uses it. NASA's Ingenuity helicopter on Mars uses it. It was built by one Swedish engineer named Daniel Stenberg in 1996. He still maintains it today, almost thirty years later. It is called curl. 42,165 stars on GitHub. The piece of software almost no one outside engineering has heard of. curl transfers data. That is all it does. It sends a request to a URL and gets the response back. HTTP, HTTPS, FTP, SMTP, and twenty-four other protocols. Every time your phone checks the weather, curl is probably running. Every time your car downloads a map update, curl is probably running. Every time you install an app, curl is probably running. It is invisible. It is everywhere. And one person built it. Here is the part that should not be possible. Daniel started curl in 1996 as a side project to fetch currency exchange rates from a website for his IRC bot. He wrote it in C. He published it for free. Thirty years later, the same project runs inside iPhones, Android phones, Volkswagens that print his name in the owner's manual, and a helicopter on Mars. Ninety-two operating systems. Twenty-eight CPU architectures. One hundred eighty thousand lines of C. Daniel has written two-thirds of every commit himself. A small team of five to ten people handles the rest. He maintained curl without pay for twenty-two years. In February 2019, wolfSSL hired him to work on curl full-time. He was forty-eight years old. He had been maintaining the plumbing of the internet as a side project for two decades. Here is what curl does not have. No VC funding. No seed round. No Series A. No marketing team. No sales team. No board of directors. No shareholders. One of the most important pieces of software on Earth has no business model. Sweden's Royal Academy of Engineering Sciences gave him a gold medal, the Polhem Prize, awarded to one engineer per year. He did not start a company. He did not raise venture capital. He did not sell to anyone. One engineer. Thirty years. Twenty billion devices. Two planets. (Link in the comments)

can’t believe they built this without Claude

A French engineer who lives quietly in Paris has spent 30 years writing software that the entire internet now runs on without knowing his name. He wrote the code that streams every YouTube video, every Netflix show, every TikTok clip. He wrote the code that runs the virtual servers underneath AWS, Google Cloud, and Microsoft Azure. He calculated more digits of pi than anyone in history. He has no Twitter. He has no marketing. He just keeps shipping. His name is Fabrice Bellard. Here is the story, because almost nobody outside the systems programming world knows what one man has built. Fabrice was born in 1972 in Grenoble, France. He studied at École Polytechnique, the top French engineering school. He never went to Silicon Valley. He never built a startup empire. He just wrote code. In 2000 he started a project called FFmpeg, an open-source multimedia framework for encoding, decoding, and streaming video. He was 28. The project did one thing nobody else had done well. It handled every video and audio format that existed, in one library, on every operating system. He led it himself for years. Today FFmpeg is the invisible engine of the internet. YouTube uses it. Netflix uses it. VLC uses it. Chrome and Firefox use parts of it. Every Android phone, every iPhone, every smart TV, every video editing tool you have ever touched runs FFmpeg somewhere underneath. If you have watched a video on a screen in the last 20 years, Fabrice's code processed it. He was not done. In 2003 he started QEMU, a machine emulator and virtualizer. He wrote it solo until version 0.7.1 in 2005. QEMU lets you run any operating system on any other operating system. It became the foundation of modern virtualization. KVM, the Linux kernel hypervisor, runs on top of QEMU. Every major cloud provider, AWS, Google Cloud, Microsoft Azure, IBM Cloud, runs virtual machines on infrastructure built around it. The Quick Emulator is the most cited piece of cloud infrastructure code on Earth. He kept going. In 2001 he won the International Obfuscated C Code Contest with a small C compiler that grew into TCC, the Tiny C Compiler. TCC can compile and boot a Linux kernel from source in under 15 seconds. In 2004 he calculated the most digits of pi ever computed at the time, using a personal desktop computer and an algorithm he derived himself called Bellard's formula. In 2011 he wrote a complete PC emulator in pure JavaScript that runs Linux in your browser, a project called JSLinux that engineers still cannot believe is real. In 2019 he released QuickJS, a small but complete JavaScript engine that fits where V8 cannot. In 2021 he released NNCP, a neural network based lossless data compressor that immediately took the lead on the Large Text Compression Benchmark. Then he turned his attention to large language models. He built TextSynth Server, a web server with a REST API for running LLMs locally. He released ts_zip and ts_sms, compression utilities that use language models to compress text and short messages at ratios traditional algorithms cannot reach. He released TSAC, a very low bitrate audio compression system. In December 2025 he released Micro QuickJS, a new JavaScript engine for microcontrollers, separate from QuickJS, designed for environments with almost no memory. Fabrice co-founded a telecom company called Amarisoft in 2012, where he serves as CTO. Amarisoft builds 4G and 5G base station software used by carriers and labs around the world. He has been running it for over a decade while continuing to ship personal projects from his own home page at bellard dot org He has no Twitter. He has no Instagram. He gives almost no interviews. His personal website is a flat list of projects with no styling, no fonts, no marketing copy. Just titles and links. A quiet French engineer who never moved to Silicon Valley wrote the code that quietly runs the internet. He is still shipping.

A Swedish developer working from a home office outside Stockholm has spent nearly 30 years maintaining a piece of software that runs on over 10 billion devices, and almost nobody who depends on it every day knows his name. His name is Daniel Stenberg. In 1996 he was building a small side project, a bot that could convert currencies inside an IRC chat channel. He needed something to pull exchange rates off the web, so he took over a tiny abandoned download tool someone in Brazil had written. He renamed it a couple of times. By 1998 he had landed on the name it still carries today. He called it curl. He never had a plan for it to become anything. In his own words he feels like a lottery winner, because he never had a long-term vision. He just wanted it to transfer data over the internet quickly and reliably, and he kept working on that one thing for the next three decades. Here is what that one thing turned into. curl now ships inside Windows, macOS, Linux, Android, and iOS. It runs in every PlayStation, every Xbox, and every Nintendo Switch. Netflix streams through it. Instagram, Spotify, and YouTube each bundle their own copy. It sits inside cars from dozens of major brands, inside printers, inside medical devices, inside set-top boxes in half a billion televisions. It has even run on hardware sent to Mars. Stenberg estimates the total is over 10 billion installations. That is more than two copies for every internet user alive. And for most of that history, it was maintained by one man. When there is a bug in curl, it does not affect one app. It affects billions of installations at once. He has said the responsibility is immense, and he means it literally. A single mistake in his code propagates to nearly every connected device on the planet. He has never run a startup. He has never chased a valuation. He does commercial support for curl to make a living and gives the software itself away for free, the way he has since the beginning. The most reliable piece of the modern internet was not built by a company with a war chest. It was built by one person who found a small problem worth solving and refused to stop solving it for 27 years. The infrastructure everything runs on is quietly held up by people you will never hear about, working alone, on projects they never expected anyone to notice.

#Chine 🇨🇳 La méthode de l'abaque mental (hand abacus), populaire en Asie, permet aux enfants de booster leur vitesse de calcul en visualisant un boulier et en simulant le mouvement des perles avec leurs doigts.RT exporttosl

🚨 BREAKING: Claude can now help you create YouTube Shorts with viral potential — completely free. Use these 8 powerful prompts to script, optimize, and grow Shorts that can reach millions of views. (🔖 Save this before everyone else does.)

🧠 Le CEO de Microsoft vient d'avertir ses propres clients : les modèles d'IA propriétaires sont des chevaux de Troie. Et oui, il parle aussi d'OpenAI. Satya Nadella a publié un essai sur son blog ce dimanche qui a provoqué une onde de choc. Sa thèse : les entreprises qui utilisent les modèles d'OpenAI et d'Anthropic paient deux fois. Une première fois en tokens. Une deuxième fois en propriété intellectuelle, sans le savoir. "Chaque requête, chaque correction, chaque interaction avec un agent IA absorbe le savoir-faire unique de l'entreprise. C'est une richesse qu'un concurrent ne pourrait jamais acheter, mais que les modèles IA obtiennent gratuitement." Nadella appelle ça le "paradoxe inverse de l'information" : plus tu utilises un modèle pour gagner en productivité, plus tu entraînes ce modèle sur ce qui te rend unique. Et ce savoir est ensuite accessible, directement ou indirectement, à tes concurrents qui utilisent le même modèle. Sa prescription : construire une "boucle d'apprentissage IA propriétaire" où l'entreprise garde le contrôle de ses données, de ses modèles fine-tunés et de son orchestration. Avec la possibilité de changer de fournisseur de modèle sans tout reconstruire. L'ironie est monumentale. Microsoft a investi 13 milliards dans OpenAI. Et son CEO dit publiquement que dépendre d'OpenAI est un risque stratégique. Mais si vous lisez entre les lignes, Nadella ne dit pas "n'utilisez pas d'IA". Il dit "utilisez-la sur Azure, avec nos outils, dans votre tenant". La mise en garde est sincère. La solution proposée profite à Microsoft. Nadella, Karp, Calacanis, Dorsey : quatre voix très différentes, le même message. Si tu ne possèdes pas ta couche IA, quelqu'un d'autre la possède. Et ce quelqu'un apprend de toi. - - - 💌Rejoignez @Disruption_IA, la newsletter quotidienne qui vous permet de suivre en 5 minutes toute l'actualité essentielle de l'IA et de la tech. Lien en bio

🚨 رئيس جوجل الحالي خرج وصدم الجميع بالتصريح ده اللي بيحدد ملامح المستقبل الكام سنة الجايين! سوندار بيتشاي (CEO شركة Google) صرّح بوضوح: "إذا لم تتعلم كيفية إدارة وتوجيه وكلاء الذكاء الاصطناعي (Agents) الآن، ستقضي عام 2027 بأكمله في محاولة اللحاق بمن بدأوا اليوم." احفظ البوست ده عندك دلوقتي عشان ترجع له قبل ما تنسى. الراجل في 30 دقيقة فقط، بيشرح ليه العباقرة وأفضل المهندسين في العالم توقفوا عن كتابة الأكواد التقليدية بإيديهم، وبقوا يركزوا في حاجة واحدة: بناء الـ Agents اللي بتدير الشغل مكانهم. المفاجأة إن غالبية الناس فاكرة إن بناء الـ Agent محتاج شهادة هندسة أو سنين دراسة.. الحقيقة إن الموضوع مش محتاج غير دليل واحد صح، وبعد ظهر يوم واحد وتطبيق جاد. لايك وفولو واحفظ البوست عشان يوصلك كل جديد!

On voit ces photos pixelisées partout sur les réseaux ! 📸 Derrière ce style, une photographe française se démarque. Refusée à la Coupe du monde, elle a décidé de placer son appareil photo devant sa télé... et le résultat est bluffant ! 🤩 Abonne-toi pour voir ➕ de portraits inspirants ! 🍀

YOUR PHONE BECOMES ALMOST USELESS THE MOMENT THE INTERNET DISAPPEARS. These 10 free apps keep maps, messaging, file transfers, medical guides, and passwords working offline. Build this kit before you need it. 1. Kiwix / kiwix.org The whole of Wikipedia, offline, in your pocket. Download it once and you carry the sum of human knowledge with no signal. It also has a full offline medical guide and a disaster-prep library built in. A nonprofit runs it, it's free, and it shows no ads and collects nothing. 2. Organic Maps / organicmaps.app Full maps of any country, downloaded to your phone, with turn-by-turn navigation that works with zero signal using just GPS. No account, no tracking, no data drain. When Google Maps spins forever, this one already knows where you are and how to get home. 3. Briar / briarproject.org A messaging app that works with no internet at all. It can send messages phone-to-phone over Bluetooth or wifi when the towers are down. Built for journalists and activists in blackouts, encrypted end to end. The one that still works when nothing else does. 4. LocalSend / localsend.org Move photos, documents, and whole folders between phones and laptops over a local connection, no internet, no account, no cloud. When you can't email yourself a file because there's no signal, this sends it straight across the room. Works on every device. 5. KeePassDX / keepassdx.com Your passwords, stored encrypted on your own phone, not in someone's cloud. When the login page won't load and the password manager needs a connection, this one opens instantly offline. Your keys stay with you, not on a server that could be down. 6. WikiMed by Kiwix / kiwix.org A complete medical encyclopedia, offline. Symptoms, first aid, what to do when you can't reach a doctor or the emergency line won't connect. The kind of information you never think about until the moment you desperately need it and there's no signal to search. 7. AntennaPod / antennapod.org Download podcasts, audiobooks, and shows ahead of time and play them with no connection. On a long blackout or a dead-zone trip, hours of things to listen to that you saved while you still had signal. Free and open, no subscription. 8. Aegis Authenticator / getaegis.app Your two-factor login codes, generated entirely on your phone with no internet needed. When you're locked out of an account because the code won't arrive, Aegis already has it. Encrypted, offline, and free. The backup nobody sets up until it's too late. 9. Offline Translator (translateLocally / RTranslator) Translate between languages with the whole model living on your phone, no connection required. Stuck somewhere you don't speak the language and there's no signal? It still works. The gap between lost and understood, closed offline. 10. FBReader or KOReader / koreader.rocks Your entire library, offline. Load it with survival guides, manuals, and every book you meant to read, and it all opens with no signal. When the screen is the only thing with power left, this keeps it worth something. The internet feels permanent right up until it isn't. Storms, blackouts, dead zones, travel. The people who prepared don't panic. They just open the kit. Build it now. You won't get to download it when you actually need it.

La Universidad de Chicago acaba de decir en voz alta lo que muchas facultades de Derecho todavía no quieren aceptar: la inteligencia artificial ya rompió el modelo tradicional de enseñanza jurídica. Hoy un alumno puede entregar una demanda, un ensayo o una investigación impecables sin haber leído, razonado ni comprendido realmente el problema. El riesgo no es solo el plagio. Es algo mucho más grave: simular que alguien sabe Derecho cuando en realidad solo aprendió a pedirle respuestas a una máquina. La respuesta de Chicago no es prohibir la IA ni rendirse ante ella. Es formar abogados capaces de pensar sin tecnología, trabajar con ella y, sobre todo, cuestionarla. Por eso vuelve al aula sin pantallas, fortalece el método socrático, exige exámenes presenciales y obliga a los estudiantes a defender oralmente sus trabajos. La lógica es brutalmente sencilla: si no puedes explicar, sostener y corregir lo que entregaste, entonces probablemente nunca fue realmente tuyo. La inteligencia artificial no va a acabar con los abogados. Va a acabar con una forma mediocre de ejercer el Derecho: memorizar, copiar formatos, repetir jurisprudencia y producir documentos sin criterio. El abogado que sobreviva no será el que escriba más rápido que una máquina, sino el que sepa detectar cuándo la máquina se equivoca, entender lo que está en juego, tomar decisiones difíciles y responder por sus consecuencias. El futuro de la abogacía no está en producir más texto. Está en tener más juicio. law.uchicago.edu/news/ai-strate…