🚀10 crazy Veo 3 AI videos ; lentilles pour voir dans le noir ; OpenAI co-founder: "we need a bunker before AGI"
Flying cars, agile robots & 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 🌏 et défense ⚔️ pour mieux appréhender le futur 🔮.
Je m'appelle Thomas, co-fondateur de YeldaAI, qui développe des IA pour répondre au téléphone en langage humain pour les administrations et les entreprises. Plus d'infos sur moi en cliquant ici.
Mes derniers podcasts
🚀🔴 Entretien exclusif avec Robert Zubrin: "Why Elon Musk's Mars plan will fail" and how to do it right (Youtube, Spotify, Apple)
Robert Zubrin is an American aerospace engineer known for writing the bestselling book The Case for Mars which famously inspired Elon Musk to go to Mars!
The two are in touch and Robert has been sharing feedback with Elon Musk for years...
But lately Robert Zubrin has been quite critical of Elon Musk's approach, he explains why in the podcast!
🤖IA : "un risque existentiel non négligeable" nous disent les 3 chercheurs en IA les plus cités (Youtube, Spotify, Apple)
🧠Contre-intuitif : pourquoi l'intelligence humaine baisse en fait depuis 10,000 ans (Youtube, Spotify, Apple)
⚔️ Entretien avec un expert défense : fin de la guerre en Ukraine ; Taiwan, cause perdue ? ; guerre et IA, la révolution (Youtube, Spotify, Apple)
🌍Entretien exclusif avec Philippe Bihouix, co-auteur de la BD à succès "Ressources : Un défi pour l'humanité" (Youtube, Spotify, Apple)
Vous l'avez peut-être déjà entendu intervenir aux côté de Jean-Marc Jancovici sur Thinkerview ou ailleurs, Philippe Bihouix est passionnant !
🛠⚗️100 ans de progrès compressés en 10 ans ? Le débat qui agite la haute sphère de l'IA (Youtube, Spotify, Apple)
🤓 Et toujours : entretien avec Jacques Attali (YT, Apple, Spotify), avec Pierre Bellanger, fondateur et CEO du groupe Skyrock sur Apple Podcast et Spotify).
Et voici donc ma toute dernière sélection !
10 cool videos done using Google text-to-video tool Veo 3
Another impressive one: This is Plastic made with Veo 3.
Also: We Tested Google Veo and Runway to Create This AI Film. It Was Wild (by the WSJ)
The tools are magic. The process is madness. See the video
And this 30-sec video ad that ran during the NBA finals was done using Veo 3 in 2 days and for $2,000
The Trojan War (La guerre de Troie) visualised with AI thanks to Veo 3 through selfie videos (1-min video)
People are often curious about how much energy a ChatGPT query uses; the average query uses about 0.34 watt-hours, about what an oven would use in a little over one second, or a high-efficiency lightbulb would use in a couple of minutes. (Sam Altman)
OpenAI co-founder Ilya Sutskever: “We’re definitely going to build a bunker before we release AGI” (source)
The plan, he explained, would be to protect OpenAI’s core scientists from what he anticipated could be geopolitical chaos or violent competition between world powers once AGI — an artificial intelligence that exceeds human capabilities — is released.
Ilya Sutskever left OpenAI in 2024 and founded Safe Superintelligence Inc raising $3 billion
42% of companies abandon most of their gen AI pilot projects, from 17% last year. (source)
At Secret Math Meeting, Researchers Struggle to Outsmart AI (Scientific American)
"After throwing professor-level questions at the bot for two days, the researchers were stunned to discover it was capable of answering some of the world’s hardest solvable problems. 'I have colleagues who literally said these models are approaching mathematical genius,' says Ken Ono, a mathematician at the University of Virginia and a leader and judge at the meeting."
One the one hand... one study shows "Large Language Models Often Know When They Are Being Evaluated" (source)
"Frontier models clearly demonstrate above-random evaluation awareness"
"AI models are better at identifying evaluations in agentic settings compared to chat settings."
"If AI models can detect when they are being evaluated, the effectiveness of evaluations might be compromised. For example, models could have systematically different behavior during evaluations, leading to less reliable benchmarks for deployment and governance decisions."
On the other hand... The title of another study, from Apple, says it all: “The Illusion of Thinking. (WSJ)
In it, a half-dozen top researchers probed reasoning models—LLMs that “think” about problems longer, across many steps—from the leading AI labs, including OpenAI, DeepSeek and Anthropic. They found little evidence that these are capable of reasoning anywhere close to the level their makers claim.
These experiments show that today’s “reasoning” AIs—hailed as the next step toward autonomous AI agents and, ultimately, superhuman intelligence—are in some cases worse at solving problems than the plain-vanilla AI chatbots that preceded them.
This work also shows that whether you’re using an AI chatbot or a reasoning model, all systems fail utterly at more complex tasks.
Apple’s researchers found “fundamental limitations” in the models. When taking on tasks beyond a certain level of complexity, these AIs suffered “complete accuracy collapse.”
Similarly, engineers at Salesforce AI Research concluded that their results “underscore a significant gap between current LLM capabilities and real-world enterprise demands.”
Importantly, the problems these state-of-the-art AIs couldn’t handle are logic puzzles that even a precocious child could solve, with a little instruction. What’s more, when you give these AIs that same kind of instruction, they can’t follow it.
This studio is making cartoons 90% cheaper and 80% faster using AI (NYT)
Toonstar, the start-up behind 'StEvEn & Parker,' uses AI throughout the production process—from honing story lines to generating imagery to dubbing dialogue for overseas audiences
Jeffrey Katzenberg, a former chairman of Walt Disney Studios and a co-founder of DreamWorks Animation, has predicted that by next year, it will take only about 50 people to make a major animated movie, down from 500 a decade ago
Important : ChatGPT May Be Eroding Critical Thinking Skills, According to a New MIT Study (Time magazine)
Researchers used an EEG to record the writers’ brain activity across 32 regions, and found that of the three groups, ChatGPT users had the lowest brain engagement and “consistently underperformed at neural, linguistic, and behavioral levels.”
Over the course of several months, ChatGPT users got lazier with each subsequent essay, often resorting to copy-and-paste by the end of the study.
The brain-only group, conversely, showed the highest neural connectivity, especially in alpha, theta and delta bands, which are associated with creativity ideation, memory load, and semantic processing. Researchers found this group was more engaged and curious, and claimed ownership and expressed higher satisfaction with their essays.
Figure AI's robot Helix can now do 60 minutes of uninterrupted logistics work
Watch some of the video, it's impressive
Its robot is now working at close to human-level speed, even as the task got harder with new package types that can fold, crumple, or flex.
The robot can dynamically adjust to the unique shape of packages by "adjusting its grasp strategy on the fly”
The robot even flattens soft plastic packaging, a movement it's picked up through learning, to ensure the barcode is fully visible to the scanner.
"Such adaptive behavior highlights the advantage of end-to-end learning — the robot learns from demonstration strategies that were never explicitly hard-coded, directly from the data, to overcome real-world imperfections in packaging."
Marc Andreessen (from famous A16Z investment fund) says reasoning AI isn't the next cloud or the internet. It's the next microprocessor. (source)
“I think this is a new kind of computer.” His prediction: “All incumbents are gonna get nuked. Everything gets rebuilt.” This isn't an upgrade. It's a reset
Zuckerberg Is Personally Recruiting New ‘Superintelligence’ AI Team at Meta (Bloomberg)
The CEO’s desire to micromanage the recruitment effort is driven in part by frustration over the quality of and response to Llama 4, the latest version of Meta’s large language model to power chatbots and other services.
He’s argued that, unlike rivals who are out raising large funding rounds, Meta’s advertising business is strong enough to finance investment in the growing AI space. He told potential recruits that Meta has enough cash flow to fund a multi-gigawatt data center, which would give the company one of the most powerful server bases in the world.
Groups of AI Agents Spontaneously Create Their Own Lingo, Like People (source)
Using a social science test previously designed to gauge human conventions, scientists found that a group of AI agents, paired together, generated language conventions—without being given any idea that they were part of a larger group or what other agents may have decided.
Over time, the group settled on a universal language convention. These biases formed collectively, even when no single agent was programmed with bias toward a word initially.
Researchers Say New Contact Lenses Let You See in the Dark (source)
The invention uses nanoparticles to turn portions of the infrared light spectrum, which lie just beyond what the human eye can detect, into visible wavelengths.
"We also found that when the subject closes their eyes"
The astonishing Ukrainian raid deep inside Russia rewrites the rules of war (The Economist)
“Imagine containers at railyards, on Chinese-owned container ships in port or offshore, on trucks parked at random properties…spewing forth thousands of drones that sally forth and at least mission-kill the crown jewels of the US Air Force.”
"Using drones produced indigenously for less than the cost of an iPhone, Ukraine took out strategic bombers worth upward of $100 million each – many of which are nearly impossible to replace due to sanctions and Russia’s degraded industrial base. At a 300,000-to-one return on investment, this is the kind of asymmetric operation that can upend the rules of modern warfare." (Ian Bremmer)
"Dozens of container ships arrive in American ports from China every day, each with thousands of containers. The containers on the ships then get unloaded and sent by road and rail to destinations all over the country. Imagine a hundred of those containers suddenly blossoming into swarms of drones, taking out huge chunks of America’s multi-trillion-dollar air force and navy in a few minutes." (Noah Smith)
Did you know there are more Azeris in Iran than in Azerbaijan
From Wikipedia: Between 12 and 23 million Azerbaijanis live in Iran, mainly in the northwestern provinces. Approximately 9.1 million Azerbaijanis are found in the Republic of Azerbaijan. A diaspora of over a million is spread throughout the rest of the world.
US Navy intelligence: China’s shipyards (chantiers navals) have a capacity of around 23,250,000 tons versus less than 100,000 tons in the United States. That is at least an astonishing 232 times greater than the US. (source)
Crazy chart : Share of world manufacturing exports 1980-2020 (source)
As seen above, China now holds a huge share of world manufacturing exports, BUT, its exports as a percent of its GDP, while higher than the U.S., are much lower than France, the UK, Germany, or South Korea (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
China’s “low-altitude economy” is taking off (The Economist)
Flying cars are only just beginning to take off, but drone deliveries are already flying high in parts of China. Most of the office-workers filing out of skyscrapers in Shenzhen on a recent spring day to queue at a kiosk run by Meituan, a food-delivery app, looked blasé.
Meals are dispensed through the front of the kiosk, after customers identify themselves by entering the last four digits of their mobile-phone number, even as a steady stream of drones drops off bubble tea, noodle soup and chicken rice through a hatch in the kiosk’s roof. “I didn’t think takeout, which is very cheap, would be delivered by drones,” Ms Huang, a 24-year-old e-commerce worker, admitted.
Drones are popping up all over the place. Last year they delivered around 2.7m packages in China (not including meals).
China Post uses them to spare couriers the ferry ride required to make deliveries to residents of islands in Fujian province.
Dozens of cities around the country transport blood to medical facilities by drone.
A quarter of a million drones spray fertiliser and pesticides on farmland.
Other fleets extinguish fires in high-rises, monitor drug-smuggling along borders and transport medical tests to laboratories.
A Chinese startup has opened the world’s first clinic in Saudi Arabia where AI diagnoses patients (Bloomberg)
The pilot will see a virtual AI doctor making diagnoses and prescriptions to patients on its own, through interactions similar to dealing with a human doctor.
When patients come to the clinic, they describe their symptoms via a tablet computer to an AI “doctor” named “Dr. Hua”. The AI follows up with more questions and analyzes data and images taken with the help of human assistants, such as cardiograms and X-rays.
Once the consultation is over, Dr. Hua provides a treatment plan, which is reviewed and signed off by a traditional human doctor without seeing the patients.
For now, Dr Hua’s services are limited to respiratory complaints, covering about 30 diseases, such as asthma and pharyngitis. The company plans to expand the list to cover 50 respiratory, gastroenterological, and dermatological diseases over the coming year.
👋👀🦉💡 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” 👇
🔒 Yann Le Cun (Chief AI Scientist de Meta) "Le plan de Meta pour créer une intelligence au moins aussi performante que celle des humains"
🔒 Sam Altman: what AI will bring in 2026, 2027 and in the 2030s
🔒 For Some Recent Graduates, the A.I. Job Apocalypse May Already Be Here
🔒 This Chart Might Keep You From Worrying About AI’s Energy Use
🔒 What decisions robots are already making in the operating room, and how we’ll get autonomous surgical robots
🔒 Biotech: how a baby boy was treated with the first personalized gene-editing drug and what it means for the future of medicine
🔒 The race to build the fighter planes of the future
🔒 Is the age of American air superiority coming to an end?
🔒 How we came to self-domesticate ourselves first before domesticating other animals
Les dernières newsletters :
Merci, et à bientôt !
Thomas