🚀"The best computer interface I’ve ever used" ; Musk Vs LeCun Twitter fight ; IA pour draguer, faire la guerre & more
Chinese gun-carrying robot dogs, Perfect product-market fit
Bonjour,
Vous recevez 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.
Je vous annonce que je suis en mesure de rétablir la gratuité de la newsletter (partie apéro) pour tous grâce au sponsoring :)
Notre premier sponsor, merci à lui, est TwinPro, la solution mobile pour vos communications professionnelles sécurisées :
Adoptée déjà par 18 millions d’utilisateurs dans le monde dans sa version Twinme & Skred, et par plusieurs milliers de nouveaux utilisateurs chaque jour
Classée parmi les 7 meilleures messageries sécurisées mondiales par le site informatique américain de référence The Verge
1ère messagerie mondiale d’origine européenne
TwinPro est la messagerie comme toutes les autres, différente de toutes les autres.
Comme les autres parce qu’elle est facile d’usage.
Différente de toutes les autres parce qu’elle n’a pas de serveur de transit de données et donc n'a pas accès aux échanges, ni quiconque.
Cette offre intéresse notamment les comex, les cabinets de conseil, les présidences et les directions générales.
Répondez-moi par simple retour d'email si vous vous voulez en savoir plus.
Et voici donc ma toute dernière sélection !
L’apéro
In case you missed it, the craziest demos OpenAI shared about their new voicebot GPT-4o
o stands for omni, because it can deal with text, audio, and video in real time.
can perform real-time audio discussions. You can interrupt without lag, and the model can hear and respond to your emotions and breathing with its own emotion (1-min video, bluffant)
2 such AIs talking to one another, one of them commenting in real time what the phone camera is filming (5-min)
OpenAI GPT-4o for education, watch that "socratic" math lesson demo by the Khan Academy, mixing shared screens and audio (3-min video)
Dad jokes 44sec ; Leading a meeting with humans 2min ; real-time translation 1min ; asking AI to talk to customer service 2 min ; AI helping friends play rock paper scissors and acting as the referee using the camera 1min ;
The new voice (and video) mode is the best computer interface I’ve ever used. It feels like AI from the movies; and it’s still a bit surprising to me that it’s real. Getting to human-level response times and expressiveness turns out to be a big change.
Talking to a computer has never felt really natural for me; now it does. As we add (optional) personalization, access to your information, the ability to take actions on your behalf, and more, I can really see an exciting future where we are able to use computers to do much more than ever before.
I don’t think anyone, including OpenAI, has a full sense of all of the implications of this shift, and what it will mean for all of us.
Best AI-experimenter and professor Ethan Mollick on the latest GPT-4 updates (source)
The Homework Apocalypse will reach its final stages. GPT-4 can do almost all the homework on Earth. And it writes much better than GPT-3.5, with a lot more style and a lot less noticeably “AI” tone. Cheating will become ubiquitous, as will universal high-end tutoring, creating an interesting time for education.
If an AI that seems to reason like a human being can see and interact and plan like a human being, then it can have influence in the human world.
New York Times' take on these updates (NYT)
the real killer feature was the way ChatGPT’s voice itself changed. One moment, it was a sing-songy soprano. The next, it shifted into a lilting contralto. It paused for effect, giggled at its own jokes and added filler phrases like “hmm” and “let’s see” for extra realism. It sounded more humanlike than some humans I know.
It also seemed to have a sense of humor. At one point during a demo, an OpenAI employee breathed in a heavy, exaggerated pant. ChatGPT heard him and responded, “Mark, you’re not a vacuum cleaner.”
a big problem with today’s A.I. voice models until now was speed. It’s hard to forget you’re talking to a robot when every answer has a three-second delay.
The most telling detail of Monday’s demo, in my view, was the way that OpenAI’s own employees have started talking to ChatGPT.
These are seasoned A.I. experts, who know full well that they are summoning statistical predictions from a neural network, not talking to a sentient being. And some of it may be showmanship. But if OpenAI’s own employees can’t resist treating ChatGPT as a human, is it any mystery whether the rest of us will?
Elon Musk will build a Gigafactory of Compute, creating a computing powerhouse four times larger than any existing AI cluster.
This AI Gigafactory will power Elon Musk's AI projects (Grok and xAI’s other projects.)
xAI, launched very recently, is now valued at almost 1/3rd of OpenAI, an 8-year-old AI company.
Yann LeCun and Elon Musk fought on Twitter (source)
Yann LeCun
Join xAI if you can stand a boss who:
- claims that what you are working on will be solved next year (no pressure).
- claims that what you are working on will kill everyone and must be stopped or paused (yay, vacation for 6 months!).
- claims to want a "maximally rigorous pursuit of the truth" but spews crazy-ass conspiracy theories on his own social platform.I like his cars, his rockets, his solar panels, and his satellite network.
I very much dislike his vengeful politics, his conspiracy theories, and his hype.
Someone: "why don’t you start your own AI company and try to do good? Staying at Meta and telling others they are wrong is a bit odd, no?"
Yann LeCun: "I'm a scientist, not a business or product person."
Elon Musk: "What “science” have you done in the past 5 years?"
Yann LeCun: "Over 80 technical papers published since January 2022. What about you?"
Elon Musk: "That’s nothing, you’re going soft. Try harder!"
Yann LeCun:
Now you're acting as if you were my boss 😂😂😂
To those who say 'these are *just* papers':
One of these papers introduced convolutional neural networks (ConvNets) in 1989.
Every single driving assistance system today uses ConvNets.
That includes MobilEye (since 2014), Nvidia, Tesla, and just about everyone else.Technological marvels don't just pop out of the vacuum.
They are built on years (sometimes decades) of scientific research that makes them possible.
Research ideas and results are shared through technical papers.
Without this sharing of scientific information, technological progress would slow to a crawl.
Wow : Nvidia pèse désormais plus en Bourse que l'intégralité du CAC 40
Nvidia sold AI chips for $22.6bn in the last quarter, versus $4.3bn in the year-ago quarter. (Quiet reminder: semiconductors are a cyclical industry, said Ben Evans)
OpenAI’s Long-Term AI Risk Team Has Disbanded (Wired)
In July last year, OpenAI announced the formation of a new research team that would prepare for the advent of supersmart artificial intelligence capable of outwitting and overpowering its creators. Ilya Sutskever, OpenAI’s chief scientist and one of the company’s cofounders, was named as the colead of this new team. OpenAI said the team would receive 20 percent of its computing power.
Now OpenAI’s “superalignment team” is no more, the company confirms. That comes after the departures of several researchers involved, Tuesday’s news that Sutskever was leaving the company, and the resignation of the team’s other colead.
Hours after Sutskever’s departure was announced on Tuesday, Jan Leike, the former DeepMind researcher who was the superalignment team’s other colead, posted on X that he had resigned (he's joining rival Anthropic AI).
The superalignment team was not the only team pondering the question of how to keep AI under control, although it was publicly positioned as the main one working on the most far-off version of that problem. The blog post announcing the superalignment team last summer stated: “Currently, we don't have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue.”
Helen Toner, a former OpenAI board member, is speaking out. (itw in Spotify podcast, op-ed in The Economist)
She claims the board wasn't informed about ChatGPT by Sam Altman but instead found out through Twitter.
She even accuses Altman of "outright lying" at times, keeping the board in the dark about OpenAI's projects and safety efforts.
Un étudiant sur trois paie 20 euros par mois pour utiliser ChatGPT et faciliter sa scolarité, selon une étude (source)
Une étude montre que GPT-4 peut analyser des états financiers avec une précision équivalente à celle des analystes professionnels (source)
«L'IA a abordé pour moi une trentaine de femmes» : demain, l’IA pourrait-elle bouleverser les rencontres sur les applis ? (Figaro, suite à une expérience menée en 2022)
«Je ne le laissais pas discuter plus de 24 heures avec les utilisatrices en question. Passé ce délai, je les prévenais qu'elles avaient discuté avec une IA», détaille Anis Ayari. « Je leur demandais ensuite si elles s’en étaient aperçues. Sur la dizaine qui a accepté de me répondre, aucune ne l’avait remarqué»,
Godfather of AI Yann LeCun reiterates: "Large language models that power generative AI products such as ChatGPT will never achieve the ability to reason and plan like humans" (Financial Times)
He said LLMs had “very limited understanding of logic . . . do not understand the physical world, do not have persistent memory, cannot reason in any reasonable definition of the term and cannot plan . . . hierarchically”.
LeCun runs a team of about 500 staff at Meta’s Fundamental AI Research (Fair) lab. They are working towards creating AI that can develop common sense and learn how the world works in similar ways to humans, in an approach known as “world modelling”.
One idea: LeCun’s team is feeding systems with hours of video and deliberately leaving out frames, then getting the AI to predict what will happen next. This is to mimic how children learn from passively observing the world around them.
Godfather of AI Geoffrey Hinton Says There's an Expert Consensus AI Will Soon Exceed Human Intelligence
"Very few of the experts are in doubt about that," Hinton told the BBC. "Almost everybody I know who is an expert on AI believes that they will exceed human intelligence — it's just a question of when."
"I think there's a chance they'll take control. And it's a significant chance — it's not like one percent, it's much more," he added. "Whether AI goes rogue and tries to take over, is something we may be able to control or we may not, we don't know."
"What I'm most concerned about is when these [AIs] can autonomously make the decision to kill people," he told the BBC, admonishing world governments for their lack of willingness to regulate this area.
Security researcher argues, unoriginally I find, that AI is not an existential threat, it's what we do with it that could be (source)
The historical reference is more interesting though:
"The thermonuclear bomb would never have been invented had it not been for an innovation in computing called the Von Neumann architecture whereby a computer could store data and run a program in memory. To successfully achieve that innovation, a single mathematical operation on a computer named MANIAC at Los Alamos National Laboratory ran non-stop for 60 days. "
"Which, then, was the weapon of mass destruction, or existential threat, the thermonuclear bomb, or the innovation in computing that brought it into existence?"
Waymo says its robotaxis are now making 50,000 paid trips every week (source)
Funny: AI bots getting stuck (The Economist)
AIs can also get tangled up. In a problem-solving experiment (...), two agents repeatedly bid each other a cheerful farewell. Even after one agent commented that “it seems like we are stuck in a loop”, they could not break free.
The list of hardware entrepreneurs who have ever built something comparable to even one of Elon’s business is very short indeed – Henry Kaiser sometimes comes to mind.
Henry John Kaiser (1882 – 1967), son of ethnic German immigrants, was an American industrialist who became known for his shipbuilding and construction projects, then later for his involvement in fostering modern American health care. Prior to World War II, Kaiser was involved in the construction industry; his company was one of those that built the Hoover Dam. He established the Kaiser Shipyards, which built Liberty ships during World War II (Wikipedia)
Agriculture and AI: A number of new pesticide-spraying methods employing AI are being commercialised, promising to cut the amount of pesticides a farmer needs to spray by a colossal 90%. (The Economist)
43 Bisons Can Help Capture CO2 From Over 40,000 Cars, Scientists Find (The Guardian)
"Bison influence grassland and forest ecosystems by grazing (paître) grasslands evenly, recycling nutrients to fertilize the soil and all of its life, dispersing seeds to enrich the ecosystem, and compacting the soil to prevent stored carbon from being released,"
China has released video footage of its gun-carrying robot dogs (see the 30-sec video)
Of course, the US military itself is experimenting with weaponized robodogs
Journalist Went To China And Drove A Dozen Electric Cars. Says "Western Automakers Are Cooked" (source)
Artificial intelligence is taking over drug development (The Economist)
The system’s predictions are tested at a large scale by means of experiments run with automated lab systems.
The results of those experiments are then used to retrain the AI and enhance its accuracy.
Recursion (...) says it can use automated laboratory robotics to conduct 2.2m experiments each week.
The US Is Cracking Down on Synthetic DNA (Wired)
The White House has issued new rules aimed at companies that manufacture synthetic DNA after years of warnings that a pathogen made with mail-order genetic material could accidentally or intentionally spark the next pandemic.
Synthesizing DNA has been possible for decades, but it’s become increasingly easier, cheaper, and faster to do so in recent years thanks to new technology that can “print” custom gene sequences.
It’s conceivable that a bad actor could make a dangerous virus from scratch by ordering its genetic building blocks and assembling them into a whole pathogen. In 2017, Canadian researchers revealed they had reconstructed the extinct horsepox virus for $100,000 using mail-order DNA, raising the possibility that the same could be done for smallpox, a deadly disease that was eradicated in 1980.
Medical AIs with human faces are on their way (The Economist)
A researcher found that people seem to find it easier to disclose their sexual history to a chatbot than to a nurse.
At least one study has found that ChatGPT’s responses to real-world health questions were preferred over those of licensed professionals both for their quality and empathy.
Perfect product-market fit : Dan Bricklin built VisiCalc in 1979, the first successful computer spreadsheet, and when he showed it to accountants it blew their minds: they could do a week’s work in an afternoon. An Apple II to run VisiCalc cost at least $12,000* adjusted for inflation, but even so, people reached for their cheque-books the moment they saw it: computer spreadsheets changed the world, for accountants. (Ben Evans)
In Portugal, a comet shot through the sky and illuminated the whole sky turquoise — and one lucky girl captured the entire extravaganza on camera.
Scientists Identify Seven Star Systems That May Be Hosting Alien Megastructures (probably not, but possible, we'll know soon) (source)
SpaceX is planning to make history this summer by enabling the first entirely commercial spacewalk.
One of the 3 Moon rover designs chosen by NASA : Flex, by company Astrolab (Ars Technica)
the most striking thing about FLEX is that astronauts are standing up rather than sitting down. This is for several reasons. Standing up provides a better view of the terrain, and it is much easier to get aboard the vehicle in a cumbersome spacesuit, Matthews said. Additionally, in the lower-gravity environment of the Moon, standing is as comfortable as sitting.
The two other teams had large traditional contractors on board Astrolab, by contrast, was small and new.
of Astrolab's 30 employees, more than 80 percent have SpaceX backgrounds
China Reaffirmed its 2030 Crewed Lunar Landing
China's crewed lunar mission involves two launches: one carrying the crewed module and the other carrying the lander. The two spacecraft will rendezvous in lunar orbit to transfer the taikonauts before the crew of two descends to the lunar surface for a short 6-hour mission.
Two Chinese astronauts just set a new spacewalking record for the country, spent about 8.5 hours working outside the Tiangong space station on May 28 (Space.com)
New trend in China: deepfake videos of Russian women (The Economist)
The women—with names like Natasha and Sofia—speak fluent Mandarin. They complain that Russian men are drunk and lazy, while praising Chinese society and technology. For a Chinese husband, the (mostly blonde) beauties say they would be delighted to cook, wash clothes and bear children.
Used to hawk products or simply to glorify China.
There are 300,000 Chinese students in the US and 800 American students in China today (The Economist)
At biotech firm Moderna, OpenAI’s GPTs Are Changing Almost Everything (WSJ)
The CEO said he’s been obsessed with ChatGPT since Christmas of 2022, and it’s one of four apps at the bottom of his iPhone, along with the apps for phone, text and internet. His goal is for employees to use it no fewer than 20 times a day, and so far, they’re keen, he said.
Of the 750 dedicated models built using ChatGPT until now, one uses years of previous research and medical knowledge to predict the optimal dose of a drug for clinical trials. Dose optimization is a huge challenge, and choosing the wrong dose can result in products being killed in the clinical-trial stage.
Another GPT combs through swaths of research to draft answers to questions from regulators—turning what used to be a weeks-long process into something that can happen in minutes, Bancel said.
Another one, on the drug manufacturing side, is helping predict the structure of new enzymes that will enable manufacturing processes with better yield and reduced waste.
Sam Altman says helpful agents are poised to become AI’s killer function (MIT Tech Review)
Altman, who was visiting Cambridge for a series of events hosted by Harvard and the venture capital firm Xfund, described the killer app for AI as a “super-competent colleague that knows absolutely everything about my whole life, every email, every conversation I’ve ever had, but doesn’t feel like an extension.” It could tackle some tasks instantly, he said, and for more complex ones it could go off and make an attempt, but come back with questions for you if it needs to.
Immigration is surging (The Economist)
The rich world is in the midst of an unprecedented migration boom. Last year 3.3m more people moved to America than left, almost four times typical levels in the 2010s. Canada took in 1.9m immigrants. Britain welcomed 1.2m people and Australia 740,000. In each country the number was greater than ever before. For Australia and Canada net migration is more than double pre-covid levels. In Britain the intake is 3.5 times that of 2019.
By the end of March, thanks to its drone attacks on Russian territory, Ukraine had destroyed around 14 percent of Russia’s oil-refining capacity... (Foreign Affairs)
... and forced the Russian government to introduce a six-month ban on gasoline exports.
One of the world’s largest oil producers is now importing petrol.
So far, Russian consumers have been largely shielded from these wholesale price increases. But in the last week of April, retail diesel prices jumped by ten percent. This lag suggests either that oil companies are earning slimmer margins, at the expense of their oligarch owners, or that the Kremlin has raised public fuel subsidies, diverting money it could have spent on the war in Ukraine.
According to some reports, the Russian government may also consider lifting restrictions on low-quality gasoline usage to prevent a fuel shortage, a move that risks damaging engines, placing further strain on an already weak military vehicle maintenance capacity and rendering void the warranties of foreign-made vehicles.
Altogether, the political, economic, and military costs are mounting for the Kremlin as the strikes on oil refineries continue.
Retired United States Navy admiral James Stavridis on impact of AI on warfare (CNN):
Logistics and maintenance: AI will help predict when and where to do maintenance, helping ensure spare parts are available at the right place at the right time
Real time advice: Whisper to commander best actions to take based on all previous battles
Swarm of AI drones: By mid century, it will be the most important aspect of warfare. Current battleships are ok for anothezer 10-15 years, vutvery vulnerable beyond that horizon
China's share of world GDP actually went down between 2021 and 2023, from 18.4% to 17%
Total workforce is actually decreasing of 7 million people per year
101 real-world gen AI use cases from the world's leading organizations (3 lines/use case, collated by Google Cloud)
Funny quote: “If we have data, let’s look at data. If all we have are opinions, let’s go with mine.” — Jim Barksdale
Partagez cette newsletter par Whatsapp en cliquant ici ❤️
Si vous appréciez cette synthèse gratuite, n’hésitez pas à prendre 3 secondes svp pour l’envoyer à ne serait-ce qu’un contact par Whatsapp en cliquant ici 🙂
Partagez la version 100% en anglais de la newsletter avec vos amis non-francophones 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 à vous abonner à la version payante pour recevoir lundi les résumés de ces articles en cliquant sur le bouton “Subscribe now” plus bas
What's your business plan in space? Find out about The great commercial takeover of low Earth orbit (MIT Tech Review)
Saudi Arabia: find out about one of the most rapid social changes anywhere in the world. (Thomas Friedman/NYT)
Famous economist Daron Acemoglu explains why he said "Don’t Believe the AI Hype" for the next 10 years (Project Syndicate)
AI Can Now Generate Entire Songs on Demand. What Does This Mean for Music as We Know It? (Author has been working with various creative computational tools for the past 15 years, both as a researcher and a producer)
The Next Frontier for Brain Implants Is Artificial Vision (Wired)
When did World War 2 begin? What about World War 3 ? (Noah Smith)
Les dernières newsletters :
L’addition ?
Cette newsletter est gratuite, si vous souhaitez m'encourager à continuer ce modeste travail de curation et de synthèse, vous pouvez prendre quelques secondes pour :
transférer cet email à un(e) ami(e) ou partager par whatsapp
et ici pour la version 100% en anglais
étoiler cet email dans votre boîte mail
cliquer sur le coeur en bas d’email
Un grand merci d'avance ! 🙏
Ici pour s’inscrire et recevoir les prochains emails si on vous a transféré celui-ci.
Merci, et à bientôt !
Thomas