🚀How a professor fought AI with AI; how a blind woman uses AI; the 3 AI bubbles; greffé des 2 bras jusqu'au épaules & more

10 Breakthrough Technologies of 2026, & more

Bonjour,

Vous recevez la version gratuite de la newsletter Parlons Futur : une fois par semaine au plus, une sélection de news, mêlant sources anglophones et francophones, résumées en bullet points sur des sujets tech 🤖, science 🔬, éco 💰, géopolitique 🌏, défense ⚔️ et espace 🚀 pour mieux appréhender le futur 🔮.

Je m’appelle Thomas, plus d’infos sur moi en cliquant ici.

🎙️ Mes derniers podcasts

• Poutine et la guerre à l’Europe : le scénario catastrophe du front atlantique

  • Entretien exceptionnel (Spotify, Apple, Youtube) avec Stéphane Audrand, consultant indépendant en risques internationaux, historien et officier de réserve dans la Marine. Il est l’auteur remarqué de l’article “Poutine et la guerre à l’Europe : le scénario du front atlantique” publié dans le Grand Continent. Au menu :

    • 2026 sera-t-elle décisive en Ukraine ?

    • L’autre menace russe complètement sous-estimée : la guerre sous-marine dans l’Atlantique

    • Le porte-avions a-t-il encore un avenir face à l’essor des missiles hypersoniques ?

    • Faut-il un second porte-avions français ? Et pourquoi faire ?

    • La Chine annexera-t-elle la Sibérie si la Russie s’effondre ?

    • La Chine parviendra-t-elle à récupérer Taiwan ? Pour qui le temps joue-t-il ?

• Vivons-nous dans une simulation ? (Youtube)

• La bulle de l’IA va-t-elle éclater ? 💥 (Youtube)

• Pourquoi Yann LeCun pense que l’Artificial General Intelligence (AGI) est un concept bidon (8 minutes, Youtube, Spotify, Apple)

• 🧠Contre-intuitif : pourquoi l’intelligence humaine baisse en fait depuis 10,000 ans (Youtube, Spotify, Apple)

• 🌍Entretien exclusif avec Philippe Bihouix, co-auteur de la BD à succès “Ressources : Un défi pour l’humanité” (Youtube, Spotify, Apple)

Et voici donc ma toute dernière sélection !

• China is now the undisputed global leader in so-called open-weight models, but its companies are struggling to make money (The Economist)

  • Open-weight models make freely available the numerical parameters learned during training (though not the underlying data or source code)

  • Chinese ones have now overtaken those from America, on Hugging Face, a popular library for such open models.

• SpaceX’s 2025 revenue estimated at $15B in 2025 with:

  • Starlink (internet access): $10.4B. with 9.2M customers, up from 4.6M in 2024

  • Launch: $4.4B

• Great example of transformative AI in action: a video of a blind woman using Meta glasses and AI to navigate the world and interact with her surroundings in real time –– giving her a new sense of independence. (video)

• Meta Lays Off Thousands of VR Workers as Zuckerberg’s Vision Fails (WSJ $)

  • The division has lost over $77 billion since its inception in 2020.

• About Yann LeCun being made by Zuckerberg to report to new employee 28-year-old Alexandr Wang (nice analysis)

  • Think about that for a second. A 65-year-old Turing Award winner. Four decades of groundbreaking research. The guy who helped BUILD this entire field.

  • Now reporting to someone whose company… labels data. (Scale AI is impressive, don’t get me wrong, but they don’t actually build AI models. They annotate training data for other companies.)

  • LeCun did an interview with the Financial Times and honestly? He chose violence. Called Wang “young” and “inexperienced.”

  • Said he has “no experience with research or how you practice research, how you do it.

  • Or what would be attractive or repulsive to a researcher.”

  • And then dropped this absolute gem: “You don’t tell a researcher what to do. You certainly don’t tell a researcher like me what to do.”

  • LeCun straight up confirmed that Meta’s team actually manipulated the Llama 4 benchmark results. Used different models on different tests to make the numbers look better.

  • Apparently Zuckerberg was furious when this came out internally.

  • LeCun says he “lost confidence in everyone who was involved” and basically sidelined the entire GenAI team.

  • LeCun has been saying for YEARS that LLMs are a “dead end.” That you can’t get to real intelligence just by predicting the next word. That we need “world models” that actually understand physical reality, not just language patterns.

  • Everyone at Meta wanted him to stop saying this publicly. Bad for the narrative, you know?

  • But LeCun refused. His exact words: “I’m not gonna change my mind because some dude thinks I’m wrong. I’m not wrong.”

  • So now he’s out. Launching his own company called AMI Labs—Advanced Machine Intelligence. They’re targeting a $3 billion valuation.

  • Building those world models he’s been talking about. Says he’ll have a “baby version” ready within a year.

• On a testé Claude Cowork, l’IA capable de tout faire (ou presque) (JDN)

  • Claude Cowork réalise le fantasme qui était porté jusqu’à présent, sans réalité, par la majorité des éditeurs d’IA du marché : disposer d’un agent universel capable de réaliser n’importe quelle tâche ou presque.

  • Le test du JDN : 1. Trier son bureau ; 2. Trier des mails et proposer des réponses ; 3. Créer un rapport à partir d’un enregistrement vocal

  • Claude Cowork est un agent IA véritablement polyvalent qui tient ses promesses.

• AI Models Are Starting to Crack High-Level Math Problems (Techcrunch)

  • “Since the release of GPT 5.2—which Somani describes as “anecdotally more skilled at mathematical reasoning than previous iterations” — the sheer volume of solved problems has become difficult to ignore, raising new questions about large language models’ ability to push the frontiers of human knowledge.”

• The Real AI Talent War Is for Plumbers and Electricians (Wired)

  • The AI boom is driving an unprecedented wave of data center construction, but there aren’t enough skilled tradespeople in the US to keep up.

• Jeep did an all AI-generated TV campaign using animated wild animals commenting on the product. At a fraction of the cost of conventional digital effects. (Youtube)

• AI is uncovering hidden geothermal energy resources (MIT Tech Review)

• Smart: This electric mining “infinity” truck doesn’t require grid charging (source)

  • It uses regenerative breaking to charge its battery while carrying its payload downhill from the mines, and this enables it to return back to the mines without requiring grid charging.

• Why we can’t have larger brains (source)

  • When women are pregnant, they need to feed two brains, which is so expensive that physically, the gut cannot mobilize enough macronutrients to keep both alive if our brains were bigger.

  • With bigger brains, we would not be able to have children — not because of the birth canal being too small, but because we would not be able to provide enough energy — making our current intelligence a physical boundary that we cannot cross due to energy limitations.

• Computer engineer gets his movement copied by a humanoid robot and...gets kicked in the balls (video) And another one of a Unitree robot doing crazy kung fu kicks

• Le tout dernier robot Helix 02 de Figure Robotics peut vider un lave-vaisselle en 4 minutes, un record (Figure.ai website, demo video)

• Boston Dynamics humanoid robot is next-level. (see the video, Wired)

  • Google DeepMind and Boston Dynamics are teaming up to integrate Gemini into a humanoid robot called Atlas.

• Tesla scraps 2 of its 4 EV models in pivot to humanoid robots and self-driving taxis as annual revenue falls for first time (Financial Times / TTSO)

• Openai’s latest fundraising came in at $40bn, larger than any IPO ever (The Economist)

• OpenAI lance une version santé de ChatGPT (source)

  • L’outil permet de synchroniser les dossiers médicaux électroniques pour des réponses personnalisées. Il a été développé avec 260 médecins sur deux ans.

  • And Anthropic does the same: Introducing Claude for Healthcare (source)

• In China, A.I. Is Finding Deadly Tumors That Doctors Might Miss (NYT$)

  • A tool for spotting pancreatic cancer in routine CT scans has had promising results, one example of how China is racing to apply A.I. to medicine’s tough problems.

  • “Several of their CT scans had raised no alarms until the cancer tumors were flagged by the A.I. tool.

  • ‘I think you can 100 percent say A.I. saved their lives,’ he said.”

• From just one night’s sleep, Stanford’s SleepFM can predict 130 diseases

• A Chinese robot performed 88% of biliary surgery on a pig using a multimodal AI model trained on surgical videos. (source)

• évaluer les candidats devient plus difficile (TTSO/Olivier Sibony)

  • avec les LLM, produire un CV ou une lettre de motivation personnalisés est instantané, or un “signal” qui ne coûte rien ne signale rien

  • En analysant près de 3 millions d’offres de travail freelance, une étude montre que les chances des meilleurs s’effondrent… et celles des nuls s’améliorent

  • Que faire ? Logiquement, les entreprises devraient miser sur les recommandations, les tests en présentiel, et la sélection après période d’essai.

  • Quant aux candidats, mon idée (Thomas) : envoyer un lien vers une vidéo de présentation et de motivation personnalisée.

• OpenAI and Sam Altman Back A Bold New Take On Fusing Humans And Machines (source)

  • Merge Labs is here with $252 million, an all-star crew and superpowers on the mind

• Surprising? In every field, elite youth performers and elite adults were almost entirely separate groups. (The Economist)

  • Around 90% of superstar adults had not been superstars as children, while only 10% of top-level kids had gone on to become exceptional adults.

  • It is not just that exceptional performance in childhood did not predict exceptional performance as an adult. The two were actually negatively correlated, says Dr Güllich.

• Lol?: “Je travaille avec une IA, elle m’appelle au téléphone quand elle a un souci” (le test JDN)

  • CallMe permet à Claude Code d’appeler son utilisateur lorsqu’il doit faire un choix ou est bloqué sur une tâche donnée.

  • Un plugin développé par la communauté open source sur GitHub qui permet de déléguer davantage de tâches en arrière-plan à l’IA. Le développeur peut alors faire ses choix au téléphone, sans revenir devant son écran.

• Lol : in the press (source):

  • Jul 2015: “Uber taxis attacked with rocks and clubs as licensed taxi drivers demonstrate against app”

  • Jan 2026: “Lyft and Uber drivers protest over rise of self-driving robotaxis”

• Google’s AI helped decode a long-unexplained handwritten annotation in a 500-year-old book (source).

• Great material for a science fiction novel?

  • Traces of Leonardo da Vinci’s DNA May Have Been Discovered on a Red Chalk Drawing Called ‘Holy Child’ (Wired)

• Students using AI killed take home exams. How a professor fought AI with AI (source)

  • with an oral final exam run by a voice agent and evaluated by a council of LLM graders.

  • > 36 students examined in 9 days
    > ~25 min avg per exam
    > $15 total all‑in (≈ $0.42/student)
    > Full transcripts, audit trail, and super actionable feedback

• “The 50x reframe” mental model for working with AI

  • Most people start working with AI by asking something along the lines of: how do I speed up what I’m already doing?

  • That question is comfortable and wrong. I find that it anchors me to existing constraints.

  • A more useful question is: What would I do if I had 50 people working on this?

  • Then work backwards.

  • The 50x reframe forces you to imagine the ideal outcome unconstrained by time or labor. Only then do you ask which parts of that hypothetical organization can be simulated with software. I now encourage our team members to think of who they would hire, what work that person would do, how they’d know if they were successful.

  • If you’ve not had the experience of hiring fifty people on a project (fair enough!), use this prompt to get you started to identify what it is that you may need:

    • I currently [describe your task/process]. Walk me through what this would look like if I had a team of 50 people dedicated to doing this comprehensively and systematically. What would each role focus on? What would the ideal output look like? Then help me identify which parts of that hypothetical team’s work could be automated or assisted by AI tools.

• «Je peins, je fais du vélo, je me gratte» : cinq ans après une première mondiale à Lyon, l’autonomie retrouvée d’un patient greffé des deux bras (Figaro)

  • Amputé des deux bras après un accident, un Islandais quitte son pays pour tenter une greffe à Lyon. En 2021, il y subit une opération inédite.

  • Aux Hospices civils de Lyon, à l’hôpital Édouard Herriot, il devient le premier patient au monde à recevoir une greffe complète des deux bras jusqu’aux épaules.

  • Cinq ans après l’opération, lors de sa visite annuelle de contrôle, les résultats continuent d’impressionner les équipes médicales. «Les bras du donneur sont vraiment les siens», rapporte France 3 Auvergne-Rhône-Alpes. Le patient a retrouvé une autonomie fonctionnelle importante. Des gestes longtemps impossibles sont redevenus accessibles.

  • Pour le professeur Aram Gazarian, cette greffe marque un tournant dans la discipline. «Il y a un avant et un après Félix»

• One irony of America’s intervention in Venezuela is that with no plans to restore democracy, it now depends on repressive Chinese technology to maintain political stability. (The Economist)

  • Chinese companies are embedded in many parts of Venezuela’s critical infrastructure.

  • Its mobile-phone networks rely on technology from Huawei and ZTE, two Chinese telecommunications giants. ZTE also developed Venezuela’s “Fatherland Card” system, which is used to track voting patterns, monitor social media and ration food.

  • Another Chinese firm provided the Maduro regime with an internet censorship system.

• The most solar punk thing in a real video: a flying drone zooming around to drop off fresh fruit (video)

• Were jobs always a fiction? (Azeem Azhar)

  • “Jobs were a fiction. Not in the sense that work wasn’t real, but in the sense that bundling tasks, identity, healthcare, social status, and income into a single institution called “employment” was a specific solution to a specific problem: how do you distribute resources and organize production when information is expensive and coordination is hard? The job was an answer to that question. It was a brilliant answer. But it was an answer to a question that is now changing.”

• Nice slide from Nat Bullard’s famous annual decarbonization presentation (source)

  

• Pendant deux semaines, de très jolies jolies influenceuses IA ont inondé TikTok de vidéos anti-Europe en Pologne. Une campagne attribuée à la Russie. (source)

• Ship-tracking data reveals China secretly mobilized thousands of civilian fishing vessels to create a 200-mile-long blockade in the East China Sea. (NYT)

• First instance ever known of a cow using a tool! (the video shared by WaPo)

• “Not dead yet!” : a hot new app is trending in China that takes the form of a countdown timer that has to be reset regularly by single people so that their friends and family will know if they die alone. (source)

Partagez cette newsletter par Whatsapp en cliquant ici ❤️

• Si vous appréciez cette première moitité de l’apéro gratuite, n’hésitez pas à prendre 3 secondes svp pour transférer cet email ou l’envoyer à ne serait-ce qu’un contact par Whatsapp en cliquant ici 🙂

• Et si on vous a partagé cet email, vous pouvez cliquer ici pour vous inscrire et ne pas manquer les prochains

👋👀🦉💡 Je vous invite maintenant à vous abonner à la version payante (l’équivalent du prix d’un café par semaine, le minium imposé par Substack) pour recevoir la semaine prochaine les résumés de ces longs articles issus des plus grands journaux (et accéder aux futures éditions payantes et archives) en cliquant sur le bouton “Upgrade to paid/Subscribe now” 👇

• 🔒 Main take-aways of Anthropic CEO Dario Amodei’s latest 30-page essay about confronting and overcoming the risks of powerful AI

• 🔒 The 10 Breakthrough Technologies of 2026 according to the MIT Tech Review

• 🔒 Que penser des robots humanoïdes : gadgets encore inutiles pour longtemps, ou révolution imminente ?

• 🔒 Anthropic’s Claude AI has a new constitution : what you need to know (2 quotes below to give a taste) …Sont perchés ?

  • “We discuss how we hope Claude will approach questions about its nature, identity, and place in the world.”

  • “Amidst such uncertainty, we care about Claude’s psychological security, sense of self, and wellbeing”

• 🔒 Why Gemini is winning

• 🔒 Who will win the trillion-dollar robotaxi race outside of China?

• 🔒 Stop calling it ‘The AI bubble’: It’s actually 3 bubbles, here they are and this is their tentative expiration date

• 🔒 Can AI companies become profitable? Lessons from GPT-5’s economics

• 🔒 Why being ethical makes economic sense after all when it comes to AI

• 🔒 The 10 most important ways the U.S. is exceptional

• 🔒 What happens if AI made the world’s economic growth explode?

Les précédentes newsletters

❤️ Transférez cette newsletter à vos amis, ou partagez-la par Whatsapp en cliquant ici.

Pour vous inscrire et ne pas manquer les prochaines éditions, vous pouvez cliquer ici.

Merci, et à bientôt !

Thomas