🚀 First synthetic cell, IA déchiffre papyrus carbonisé et déniche footballeurs, Top 10 Emerging Tech of 2026, robot footballeur &+

Why lasers won’t stop drones

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.

• Vivons-nous dans une simulation ? (Youtube)

• 🌍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 !

• Alex Karp, le CEO de Palantir, a poussé un sacré coup de gueule cette semaine (7-min version, 19-min version)

  • He said businesses depend on a handful of frontier labs, pay for tokens that create little value, and risk handing over the data that makes them valuable.

  • He said this wouldn’t last.

  • Here’s what he suggests everyone should do instead: run good enough models on your own data while keeping control of the weights, the data, and the compute.

  • This is obviously self-serving as it assumes a service layer his company can offer, but there seems to be more than a grain of truth to it!

• Des chercheurs ont déchiffré un rouleau de papyrus carbonisé vieux de 2.000 ans grâce à l’IA après son incinération lors de l’éruption du Vésuve (source)

• Scientists say they have built a cell from scratch for the first time (CNN)

  • Scientists say they have built a cell from scratch for the first time that can feed, grow and replicate like a natural cell. This breakthrough in synthetic biology could usher in an era of made-to-order organisms that function like living machines.

• Mark Zuckerberg Just Got Rather Badly Humiliated (Financial Times)

  • Google informed Meta earlier this year that it was cutting Meta off from the Gemini AI capacity it needed

  • Meta chose Gemini over its in-house Llama open-source models simply because it performed much better

  • Despite committing over half a trillion dollars over the next two years, the company’s flailing AI efforts have been plagued by rock bottom morale, infighting, high-level executive departures, and related PR crises.

• Nvidia says AI’s water challenge is largely solved (Axios)

  • Nvidia’s coolant — a recirculated liquid mixture that includes water and propylene glycol, similar to automotive antifreeze — can run at 45°C

  • Because the liquid can operate at higher temperatures than previous systems, data centers may be able to rely less on chilling equipment that uses large amounts of energy or water — or even eliminate it altogether.

• Is China AI really cheaper than American AI? (The Economist)

  • DeepSeek charges just $0.87 per 1m output tokens for its v4 model, whereas Anthropic charges $50 for the same on Fable 5.

  • A study updated this month by Du Zheng of Georgia Tech and coauthors shows that given the same tasks, a DeepSeek model used 23 times more tokens than its Openai rival to achieve basically the same result.

  • Because of these large differences in efficiency, the correct way to compare models is not price per token but the total cost of all the tokens used.

  • Using this metric, on a benchmark designed to test software engineering, glm 5.2 (by Beijing-based lab called Zhipu, or Z.ai,) ended up costing more than competing systems from Anthropic and OpenAI.

• Apple said that it will not be providing the new AI features in the EU because of the Digital Markets Act’s competition requirements. (Ben Evans)

  • Apple says that it went to the EU and said “we can build an intermediate brokerage and security layer, so that third party assistants could do what Siri will do, but that would be 18 months of work, so you’re going to need to tell us if you would approve it before we do that work” - and Apple’s story is that the EU said, “We’re not going to tell you. Build it and then we’ll make a decision”.

  • The Digital Markets Act has very wide scope for interpretation and it’s a matter of opinion whether X or Y product implementation would follow the law, so the tech company goes to the EU and says “will you agree that this follows the law?” and the EU says “we’re not going to tell you - ship it, and then we’ll decide, and if you guess wrong we’ll fine you”. It’s very hard to see this as a good faith attempt at regulating a complex and fast-moving sector.

• There used to be a joke that an ‘AI scientist’ is a statistician who lives in San Francisco

• Anthropic Blames Dystopian Sci-Fi for Training AI Models to Act ‘Evil’ (Ars Technica)

  • “Those with an interest in the concept of AI alignment may remember when Anthropic claimed its Opus 4 model resorted to blackmail to stay online in a theoretical testing scenario last year.

  • Now, Anthropic says it thinks this ‘misalignment’ was primarily the result of training on ‘internet text that portrays AI as evil and interested in self-preservation.’”

• If Openai and Anthropic each gave 3% of their equity to the Americans, and their combined value were to rise ten- or 20-fold, Americans would “only” get a yearly payout of a few hundred dollars per year in dividends (The Economist)

  • Assuming it would feed a fund earning a healthy 10% a year, and after 10 years paying out 4% annually, a rule of thumb for preserving a fund in perpetuity

  • This is not enough to make anyone rich…

• China Aims AI at Predicting Who Could Pose a Political Risk (NYT)

  • the Chinese company’s researchers were working to develop behavioral profiles of people based on telecommunications, social media and location data.

  • A.I. models were used to classify people and to “detect harmful information,”

  • the research team was doing more than just documenting behavioral patterns. They were trying to predict what citizens might do next and with whom,”

  • “This is what happens when mass surveillance meets A.I.,”

• China is innovative. Its economy is a mess: some crazy stats (The Economist)

  • local-government borrowing in China is at 43% of GDP. The comparable figure in America is 12%.

  • The share of industrial firms generating losses has shot to a record high of around 32% in April, up from 10% in 2011 and above the previous peak during the Asian financial crisis in 1998.

  • Is China’s industrial policy efficient ? Well:

    • a trio of IMF economists calculated last year, China’s “total factor productivity” (which captures how efficiently both capital and workers are used) was 1.2% lower than it would have been in the absence of industrial policy over the past decade or so.

    • GDP was 2% lower, equivalent to forgoing around $400bn in value added a year.

• Don’t restrict Chinese biotech (The Economist)

  • Chinese firms ran nearly 33% of the world’s clinical trials last year, up from just 6% a decade earlier.

  • China is now the world’s second-largest source of new drugs, behind only America itself.

  • In 2025 nearly half of licensing deals worth $50m or more were struck with Chinese firms, up from none in 2020.

  • In some categories, such as antibody-drug conjugates, a promising class of cancer treatments, Chinese firms accounted for almost all the licensing.

• AI and doctors : “And we already know from data that A.I. is better at medical diagnosis than doctors under many circumstances. And controlled experiments show patients prefer talking to A.I. over doctors because it has higher perceived empathy” (NYT)

  • said Ethan Mollick, professor at the Wharton School of the University of Pennsylvania and the author of “Co-Intelligence” and the forthcoming “Co-Existence”

• AI Is Designing Radio Chips That Humans Couldn’t Even Imagine (IEEE Spectrum)

  • “Some of the...chips look more like modern art than circuit layouts. Yet in many cases, the physical prototypes bested state-of-the art circuits in terms of performance. The real achievement, however, is that it took the AI orders of magnitude less time to conceive a working design than it would a human designer.”

• Economists are actually quite conservative about AI impact on growth (The Economist)

  • Even under a scenario where AI progress is “rapid” by 2030—meaning that AI can compete with or surpass the brainiest humans—the median academic economist expects American GDP growth of just 3.5% in 2050 (compared with 5.3% for AI researchers).

  • Only 11% of leading economists agree that the use of AI over the next decade “will lead to a substantial increase in the unemployment rates in advanced countries”, according to a survey by the University of Chicago.

• A Startup Claims It Broke Through a Bottleneck That’s Holding Back LLMs (MIT Tech Review)

  • “According to Subquadratic, it has developed a new kind of LLM, called SubQ, that is faster and cheaper and uses a lot less energy than any other model on the market.

  • The company also claims that SubQ is able to process up to 12 times as much text at once than most other models, allowing it to carry out a range of data-heavy tasks, such as analyzing hundreds of documents or entire code bases.”

• With Starfall, SpaceX eyes an edge in global cargo delivery from orbit (Ars Technica)

  • A SpaceX Falcon 9 rocket lifted off in late June to test a new reentry vehicle designed to deliver cargo anywhere in the world from low-Earth orbit.

  • The Starfall vehicle is cylindrical in shape, with a diameter of 3.1 meters and a height of 0.75 meters) Starfall weighs approximately 2.1 metric tons with a capacity for about 1 metric ton of payload.

  • Starfall will “enable point-to-point delivery of critical cargo through space on rapid timelines” and provide access to space for commercial in-space manufacturing, a nascent market that, so far, is largely geared toward pharmaceuticals.

  • So who might use something like Starfall? The US military is one obvious answer.

• SpaceX said it had performed some 300,000 maneuvers to avoid collisions in 2025. (NYT)

• Video of Boston Dynamics’ Atlas robot playing football (39-sec video)

• And a Chinese robot kicking the ball So Hard It Smashes Hole in Wall (28-sec video)

  

• Humanoid Robot Preparing to Climb Mount Everest (2-min video)

  • a modified Unitree G1 unit has recently completed a trek to the top of Ecuador’s Chimborazo volcano, at more than 6,000 meters.

  • the robot walked autonomously along sections of the 16-hour trek to the summit with less than a 30-degree incline, but was carried for the rest.

• SoftBank annonce un investissement record de 75 milliards d’euros dans des infrastructures liées à l’intelligence artificielle en France (source)

  • “Ce sera l’investissement le plus important en Europe dans les infrastructures liées à l’intelligence artificielle: 75 milliards d’euros au total, dont 45 milliards d’ici à 2031, dans les Hauts-de-France”, a détaillé Masayoshi Son

  • “Le premier facteur positif, c’est l’énergie”, a-t-il déclaré, évoquant le fait que “le pays est producteur et exportateur d’énergie” ce qui “est absolument décisif pour des investissements d’infrastructure dans l’intelligence artificielle”.

  • Ensuite, “la France est extrêmement impliquée dans des domaines cruciaux, tels que la tech en général et l’IA en particulier”

  • et enfin, Masayoshi Son a loué “l’excellent système éducatif” et le “large vivier de talents disponibles, en particulier dans l’ingénierie et la tech”.

  • Mais attention alerte Gilles Babinet : “Un autre aspect qui mérite de l’attention, c’est de connaître la nature des clients de ces datacenters : s’il s’agit d’Hyperscalers ou même de sociétés comme CoreWeave ou Crusoe, cela signifie que la valeur ajoutée française s’arrêtera à la fourniture d’énergie et d’agents de gardiennage.”

• Google DeepMind CEO Demis Hassabis said at Google’s developer conference last week that humanity is standing in the “foothills of the singularity” (Axios)

  • and that society has only a few years left to prepare for AGI.

  • Hassabis said his prediction that AGI could arrive in four years — or even sooner — reflects growing confidence that the industry has found the right technical path.

• Premier long-métrage où une IA tient l’histoire et les personnages du début à la fin. (source, partagée par Futur Proche)

  • Hell Grind, 95 minutes de fantasy violente entre Mortal Kombat et Marvel, a été produit en deux semaines par quinze personnes pour 500 000 dollars, dont 400 000 partis dans le seul calcul des images.

  • Chaque plan de quinze secondes a réclamé un prompt d’environ trois mille mots.

• Les Large Language Models sont-ils capables d’émotions ? (source, partagée par Futur Proche)

  • Chris Olah, chercheur en interprétabilité et cofondateur d’Anthropic, en réponse à l’encyclique du pape Léon XIV :

  • “Je dirige une équipe de recherche qui étudie la structure interne de ces modèles, ce qui s’y passe vraiment. Et je serai honnête, nous y trouvons sans cesse des choses mystérieuses, voire troublantes. Nous trouvons des structures qui reflètent les résultats des neurosciences humaines. Nous trouvons des preuves d’introspection. Nous trouvons des états internes qui imitent, sur le plan fonctionnel, la joie, la satisfaction, la peur, le chagrin et le malaise.”

• “Notre robot araignée grimpe aux parois des navires pour effectuer des tâches dangereuses (JDN)

  • Actuellement, il effectue des opérations de soudage. Et à l’avenir, il pourra aussi faire de la peinture, des contrôles non destructifs, ou encore du décapage par projection d’abrasifs

  • “En réalité, le robot araignée est bien plus complexe qu’un humanoïde”

• La vie presque partout : millions of bacteria inhabit fog droplets (gouttelettes de brouillard) (source)

  • “When you take all of the droplets together, the concentration of bacteria is the same as in the ocean,”

  • Some thrive more than others. The population of one bacteria called Methylobacteria, known for devouring simple carbon compounds including pollutants like formaldehyde, increased after fog events.

• At 92, she’s as fit as someone half her age (Washington Post)

  • “What we are learning is that, definitely, exercise training is a major thing we should focus on to change the aging process,”

  • For about 15 years, she didn’t race or formally exercise much, until, at about age 45, with her children grown, she decided to start sprinting again.

  • The bulk of her workouts involve high-intensity interval training, meaning the exercise effort is draining, but the duration short. Her training sessions rarely last more than an hour, and she almost never goes for long runs or completes other sustained aerobic exercise. She does walk outside almost every day and occasionally performs light resistance exercise.

  • Her overall aerobic fitness resembled that of athletic women almost half her age, and her ability to take in oxygen and start passing it along to her muscles was even better, comparable to what’s seen in active 30-year-olds.

  • Just as impressive, her muscles’ mitochondria — the tiny organelles inside cells that absorb oxygen and transform it into energy — were exceptionally robust, functioning like those of 20-year-olds.

• NASA Rover Finds “Complex Organic Matter” on Mars (source)

  • The scientists found that samples from the rocks that Perseverance originally found in 2024 contained macromolecular carbon (MMC), large networks of carbon atoms commonly found in Earth rocks containing fossilized biological carbon and meteorites.

  • Scientists were baffled how close to the surface — mere microns away — the MMCs were preserved, considering the extreme environment.

• Parlons Passé

  • Le roi du Royaume-Uni Charles III descend directement de Guillaume le Conquérant, de Charlemagne et Clovis...

    • ...mais aussi incroyable que cela puisse paraître, statistiquement, toute personne aujourd’hui qui a une ascendance européenne ou méditerranéenne, descend de TOUS les êtres humains qui ont vécu jusqu’en l’an 1000 environ et ont eu une descendance jusqu’à ce jour. (source)

      • Donc toute personne aujourd’hui qui a une ascendance européenne descend aussi de Clovis, Charlemagne, etc.

      • Mais la probabilité que quelqu’un au hasard en Europe en 1400 soit un de vos ancêtres, si vous avez une ascendance européenne, tombe à quasi 0% (contre quasi 100% en l’an Mille donc !)

      • En deux mots, si vos ancêtres étaient tous distincts, vous en auriez une bonne dizaine de milliards en l’an mille, mais il n’y avait que quelques dizaines de millions d’Européens, donc mathématiquement, TOUS ces anciens Européens, s’ils ont eu une descendance à ce jour, ont 99,999999999999...% de chances de compter parmi vos ancêtres !

      • Mais, contre-intuitif, ils n’ont chacun quasi 0% de chances de vous avoir légué de l’ADN, car l’ADN se lègue par blocs et on n’hérite que de la moitié de l’ADN d’un de ses deux parents. La quasi totalité de nos ancêtres ne nous ont laissé aucune trace génétique, dingue !

  • Il y a environ 900 000 ans, l’humanité s’est retrouvée réduite à environ 1 280 individus reproducteurs (source)

    • Cette crise démographique, qui a balayé 98,7 % de la population humaine de l’époque, s’explique principalement par des facteurs environnementaux catastrophiques

    • La majorité des paléoanthropologues et des biologistes pensent aujourd’hui que si ces 1 000 individus avaient disparu, une intelligence de type humaine n’aurait probablement plus jamais émergé, car

      • 1. La niche de “l’hyper-intelligence” n’est pas obligatoire

      • 2. Le réservoir d’hominidés était vide (ou presque)

      • 3. La contingence historique (l’effet papillon) : L’apparition de l’intelligence humaine n’est pas une ligne droite. C’est le résultat d’accidents de l’histoire.

    • Si vous supprimez les 1 000 survivants, vous cassez la chaîne. Les autres hominidés restants n’avaient pas nécessairement les bases génétiques ou les conditions environnementales pour prendre le même chemin évolutif. L’Afrique serait probablement restée un continent dominé par de grands singes arboricoles ou des prédateurs classiques.

    • En résumé
      L’apparition d’une espèce capable de se poser des questions sur ses origines n’était pas inscrite dans les gènes de la Terre. Nous ne sommes pas le sommet inévitable de l’évolution, mais les descendants des gagnants d’une loterie cosmique et climatique ultra-sélective.

Partagez cette newsletter par Whatsapp en cliquant ici ❤️

• Si vous appréciez cette newsletter, 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

---

Could A.I. Find Brazil’s Next Pelé? (NYT)

• Drones and AI-powered apps are used to monitor player performance during training sessions.

• Their speed, control and footwork were all being carefully evaluated — but not by a veteran scout. Not even by a human being.

• Instead, the Brazilian athletes were being assessed by a mobile app powered by artificial intelligence, part of a handful of new tools promising to revolutionize the way that talent is discovered in the soccer-mad nation.

• now, powerful A.I. technology could identify emerging talent faster and better than the human eye alone. It also could reach more aspiring athletes across Brazil’s vast territory, giving even those in remote reaches of the country, where few scouts might venture, a chance to be spotted.

• Most A.I. scouting platforms work by analyzing videos that users upload or drills they record directly into the apps. Weighing a vast array of abilities, from speed to ball control, they spit out a score and add athletes to a database. There, human agents seeking talent can find them, or the apps can pitch athletes directly to specific clubs.

• On a battered phone, Mr. Barossi showed off his own metrics. He worried because he is shorter than many players his age, he said. But, after doing drills through the app every day, he had climbed into the top 30 nationwide in his age group. “I’m always kicking the ball around and recording it,” he said.

---

👋👀🦉💡 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” 👇

• 🔒 Why lasers won’t be able to stop drones

• 🔒 Top 10 Emerging Technologies of 2026

• 🔒 AI Isn’t Conscious. Here’s why we aren’t either

• 🔒 2028: Two scenarios for global AI leadership (Anthropic)

• 🔒 What you need to know: Anthropic Urges Global Pause in AI Development, Flags ‘Self-Improvement’ Risk

• 🔒 Les fourmis sont-elles douées de conscience ?

• 🔒 Why it’s impossible to predict which jobs AI will destroy

• 🔒 Why ‘Can a machine do this job?’ is the wrong question

• 🔒 The revolution in AI reforestation that is underway

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