🚀 Tesla does something Bill Gates said wasn’t possible, AI lets you talk to younger self, hibernate to visit the future & more
Disney AI de-ages you in seconds, AI can tell male from female eyes, & more
Bonjour,
Vous recevez la newsletter Parlons Futur : une fois par semaine au plus, une sélection de news résumées en bullet points sur des sujets tech 🤖, science 🔬, éco 💰, géopolitique 🌏 et défense ⚔️ pour mieux appréhender le futur 🔮.
Je m'appelle Thomas, plus d'infos sur moi en bas d'email.
Voici donc ma dernière sélection !
L’apéro
The European Commission's foreign aid department threw a €387,000 24-hour metaverse "beach party" full of "music and fun" (source)
less than 10 guests showed up and stayed less than an hour
San Francisco Just Reversed Its Killer Robot Plan (source)
More than a week on, those same legislators have rolled back their decision, sending back the plans to a committee for further review.
On December 5, a protest took place outside San Francisco City Hall, while at least one supervisor who initially approved the decision later said they regretted their choice.
“I regret it. I’ve grown increasingly uncomfortable with our vote & the precedent it sets for other cities without as strong a commitment to police accountability. I do not think making state violence more remote, distanced, & less human is a step forward.”
Exemple de projet artistique utilisant l'IA générative : "Le faux documentaire Netflix The Island of The Simulated Wolves" (source)
Imaginez une île sauvage remplie d’animaux synthétiques et de glitches (un terme anglais désignant des bugs graphiques) générée par IA. Cette île est le cadre du documentaire The Island of The Simulated Wolves, raconté par Samuel L. Jackson et diffusé sur Netflix. Mais n'espérez pas le voir sur la plateforme de streaming, il s'agit d'une création parodique de l'artiste Vanessa Rosa.
Dans cette vidéo expérimentale, tout a été généré par IA : la vidéo est issue d’un mariage entre Stable Diffusion, un algorithme permettant de générer des images, et l'outil EbSynth, qui anime des images fixes. La voix typique de Samuel L. Jackson est, quant à elle, générée par l’outil de synthèse vocal Uberduck. Le tout donne une petite pépite hallucinante.
ChatGPT a été lancé avec toutes sortes de filtres pour éviter les abus et détournements, mais des utilisateurs parviennent tout de même à le faire dérailler : "Seeing people trick ChatGPT into getting around the restrictions OpenAI placed on usage is like watching an Asimov novel come to life."
Source, avec d'autres exemples
More examples of ways to bypass OpenAI’s content filters: by asking the model to only pretend to be evil, pretend to break into someone’s house (MIT Tech Review)
Cool example of the application of GPT-3 in real life:
Danny mentors a young man with poor literacy skills who is starting a landscaping business. He struggles to communicate with clients in a professional manner. So, Danny created a GPT3-powered Gmail account to which he sends a message. It responds with the text to send to the client.
OpenAI liked it so much they are collaborating to make a tool specifically for these kinds of use cases AND it was made with no-code applications. The landscaper just secured his first contract worth £220k. (source) How it works here.
Amazing: Someone asked ChatGPT a one-line question to get prompts instantly for image-making AI tools, then fed them into one to get images in seconds, see one example below
Sam Altman, co-founder & CEO of OpenAI said in September "I don’t think we’ll still be doing prompt engineering in five years", well, just 3 months later, it doesn't seem to be necessary indeed
Someone Taught ChatGPT to Invent a Language (source)
It understands subordinate clauses (meaning it understands at least one level of recursive grammar, which Chomsky thinks is the basis for all human grammar)
it does generalize to some degree, and when asked to answer questions in the made-up language, on which is barely trained, it answers correctly most of the time.
"Ipop gloop glog bluba trom plog plopa slurpi" - This sentence means "The happy slime sees the tree drink the water with its mouth" in Glorp.
AskAlfred is a chrome extension that gives you AI tool GPT’s second opinion alongside your google searches : askalfred.co
Elon Musk demande à Sam Altman, CEO d'OpenAI, combien leur coûte une réponse via ChatGPT (gratuit pour les utilisateur, lancé la semaine dernière et comptant déjà plus d'un millions d'utilisateurs) : "average is probably single-digits cents per chat; trying to figure out more precisely and also how we can optimize it" (source)
One theory: Someone speculated that the gap between AI and humans will accelerate not only because AI keeps getting better, but also because people using AI will get less smart. Similar to how we cannot drive in our own cities without GPS. (source)
Cool example of AI fails (more examples here)
Dog Buttons are a growth industry: dogs can learn to communicate by pushing different coloured buttons to ask for things. (more here: The best dog buttons)
Another AI tool, just in time for Xmas: storytimes.ai, feature your kid in a children's book with AI
40% of global shipping involves moving fossil and other fuels (oil, gas, wood pellets) around. More renewables (solar, wind, nuclear, geo), means fewer ships. (source)
Scientists finally know why people get more colds and flu in winter (CNN)
In the study, published in The Journal of Allergy and Clinical Immunology today, the researchers found that just a12°C drop in temperatures can kill almost 50% of the cells that fight bacteria and viruses inside the nostrils.
"Cold air is associated with increased viral infection because you’ve essentially lost half of your immunity just by that small drop in temperature," Benjamin Bleier, coauthor and Harvard Medical School rhinologist, told CNN.
There's a surprisingly simple way for us to keep the cold and the flu away during the winter: keeping our nasal cavities from getting too cold by wearing a mask. (ou un cache-nez)
A machine learning model can tell male from female retinas. No-one knew there was a difference. "Machine learning can give you infinite interns, but it can also give you one intern with infinite patience - that can look at 85k photos of retinas and spot a pattern no-one knew was there." said tech analyst Ben Evans (from Nature, 2021)
À votre bon coeur ❤️
Si vous appréciez cette synthèse gratuite, n’hésitez pas à prendre 3 secondes svp pour l’envoyer à ne serait-ce qu’un contact 🙂
Et si on vous a partagé cet email, vous pouvez cliquer ici pour vous inscrire et ne pas manquer les prochains
À table !
Disney made a movie-quality AI tool that makes actors look younger or older in seconds (source; 6-min explainer by Disney)
Making actors look older or younger in movies used to be a huge deal, the amount of postproduction work to achieve realistic results was immense back in the day,
but now, researchers from Disney have revealed FRAN, a new artificial intelligence tool that can convincingly age or de-age an actor in a fraction of the time.
Someone trained an AI chatbot on her childhood diaries so she could engage in real-time dialogue with her "inner child" (MIT tech Review)
Michelle Huang, a coder and artist, wanted to simulate having conversations with her younger self, so she fed entries from her childhood diaries to the chatbot and had it reply to her questions. The results are really touching.
She told her story on Twitter
"conversing with "younger Michelle" reminded me of the parts of myself that have stayed constant through the years, but also of the parts that i forgot or buried as life went on, it was like holding a mirror to an unapologetic, more earnest, and pure version of my own essence"
"these interactions really elucidated the healing potential of this medium: of being able to send love back into the past, as well as receive love back from a younger self"
A good summary of where self-driving is at today (François Chollet, ingénieur français chez Google)
$100B has been invested in the field overall over the past 10 years — roughly half of the inflation-adjusted cost of the Apollo program.
And we’re now just starting to see fully driverless cars able to handle a controlled subset of all possible driving situations.
You can ride in one in SF from GM-backed Cruise (in private-access beta) or in SF or Phoenix from Alphabet-owned Waymo (in public access). Other players tend to be far behind on the generality curve.
Crucially, these results were not achieved via some kind of “just add more data and scale up the deep learning model” near-free lunch.
Deep learning only represents a subset of the system. These results come from years of work-intensive engineering that went into crafting complex systems that encompass millions of lines of human-written code. And we’re still very, very far from generalizing to all locales or all situations.
Watch Tesla Semi Truck do something Bill Gates said wasn’t possible
Tesla has released a timelapse video of its Tesla Semi electric truck (camion semi-remorque) completing a 800-km trip with a full load on a single charge – something Bill Gates and Daimler said wasn’t possible just a few years ago. (see the 2-min timelapse video)
In a blog post, Gates argued that all-electric semi-trucks like the Tesla Semi would “probably never” work because batteries would be too heavy:
The problem is that batteries are big and heavy. The more weight you’re trying to move, the more batteries you need to power the vehicle. But the more batteries you use, the more weight you add—and the more power you need. Even with big breakthroughs in battery technology, electric vehicles will probably never be a practical solution for things like 18-wheelers, cargo ships, and passenger jets. Electricity works when you need to cover short distances, but we need a different solution for heavy, long-haul vehicles.
Daimler’s head of trucks, Martin Daum, who said that Tesla’s 500-mile range broke the laws of physics:
If Tesla really delivers on this promise, we’ll obviously buy two trucks — one to take apart and one to test because if that happens, something has passed us by. But for now, the same laws of physics apply in Germany and in California.
Earlier last week, Tesla proved them wrong, Elon Musk said: "There are some people out there that say it can't be done. I don't know who might say that, but I've heard rumors. So we... just did it".
the video shows a continuous drive with a single break without charging over 800km.
The truck also had to go climb quite a bit. On a flatter route, it could likely achieve a range closer to 965 km.
The full 35-min Tesla Semi Delivery Event ; the 8-min video summary.
Musk’s Neuralink faces federal probe, employee backlash over animal tests (Reuters)
I covered Neuralink's last presentation quite extensively last week here, but there has been a new development
In all, the company has killed about 1,500 animals, including more than 280 sheep, pigs and monkeys, following experiments since 2018, according to records reviewed by Reuters and sources with direct knowledge of the company’s animal-testing operations.
The total number of animal deaths does not necessarily indicate that Neuralink is violating regulations or standard research practices. Many companies routinely use animals in experiments to advance human health care
But current and former Neuralink employees say the number of animal deaths is higher than it needs to be for reasons related to Musk’s demands to speed research. Through company discussions and documents spanning several years, along with employee interviews, Reuters identified 4 experiments involving 86 pigs and 2 monkeys that were marred in recent years by human errors.
The mistakes weakened the experiments’ research value and required the tests to be repeated, leading to more animals being killed, three of the current and former staffers said.
On several occasions over the years, Musk has told employees to imagine they had a bomb strapped to their heads in an effort to get them to move faster, according to three sources who repeatedly heard the comment.
The mistakes leading to unnecessary animal deaths included one instance in 2021, when 25 out of 60 pigs in a study had devices that were the wrong size implanted in their heads, an error that could have been avoided with more preparation, according to a person with knowledge of the situation and company documents and communications reviewed by Reuters.
Neuralink executives have said publicly that the company tests animals only when it has exhausted other research options, but documents and company messages suggest otherwise.
Recent science may make it possible for humans to hibernate so as to explore the solar system (Wired, 5900+ mots réduits à 700 mots)
Hibernation may be the only remotely attainable form of long space (and time) travel.
In 2013, Bradford persuaded NASA's Innovative Advanced Concepts program to fund a project assessing the feasibility of “human torpor.” His successful pitch centered on the potential weight savings: He estimated that if astronauts could be kept frigid for the bulk of their trip to Mars, the mass of their life-support resources could be cut by as much as 60%. Bradford also hypothesized that torpor could help astronauts fend off a number of serious health hazards, ranging from radiation to the psychological perils of extreme boredom and isolation.
There are many species that go torpid every winter, drifting into an unconscious state that drastically squelches their bodies' cravings for food and air. When they rapidly whirr back to life in spring, these creatures show no signs of suffering from muscle atrophy, malnourishment, or other ailments that might be expected to stem from lengthy spells of idleness.
The Arctic ground squirrel is the most extreme hibernator on the planet, it hibernates for up to 8 months of the year. During that span, the animal's internal temperature falls to below -2.0°C, literally as cold as ice. Its brain waves become so faint that they're nearly impossible to detect, and its heart beats as little as once per minute. Yet the squirrel remains very much alive. And when spring comes, it can elevate its temperature back to 37°C in a couple of hours.
Since so many hibernators are our close genomic cousins, there is good reason to believe that we can tweak our brains and bodies to mimic what they do.
In March 2018, NASA asked biologists: if provided with sufficient backing, they could help humans achieve at least some level of true hibernation in the next 10 to 15 years—a timeline that dovetailed nicely with NASA's plans to send humans to Mars in the late 2030s or early 2040s.
In July 2018, SpaceWorks published a report that asserts that, based on the current pace of research, NASA could begin testing hibernation technologies on human subjects as early as 2026. Judging by investments that NASA has initiated in recent months, the agency seems intent on making that happen.
One of the great unknowns about the mission to Mars, for example, is whether humans can endure the ravages of galactic cosmic rays, the remnants of the Milky Way's celestial violence. Once a spacecraft travels beyond Earth's protective magnetosphere—which orbiting craft like the International Space Station stay well within—there's no real way to dodge these cancer-causing particles, and scientists have yet to find a malleable, lightweight material that can shield against them. But if human cells can be made less active, they may develop significant resistance to radiation.
In a 1972 experiment, for example, scientists found that ground squirrels that were irradiated while hibernating had a much higher survival rate than their fully conscious peers.
Callaway doubts that when those astronauts sleep they'll ever get as cold as the Arctic ground squirrel or have metabolisms as low. But he notes that bears are pretty effective hibernators too, and they reduce their internal temperature by only a few degrees while snoozing through a winter. “In this decade,” he says, “we can replicate that.”
But if hibernation does indeed become a realistic option for humans, even those of us in decent shape may find it tempting. Induced torpor seems to offer a roundabout path to realizing at least a couple of transhumanist dreams. Like life extension, perhaps—provided you're not purely bent on extending your conscious life.
As American scientist Raymond J. Hock noted in 1960, hibernation really does seem to offer a fountain of youth. Earlier this year, for example, a team at UCLA found that yellow-bellied marmots, which hibernate for as much as 66% of every year, possess much more robust genetic material than might be anticipated based on their chronological ages. “The molecular and physiological responses required for an individual to successfully hibernate may prevent aging,” the researchers wrote in Nature.
A biological fast-forward button that would allow someone to skip months or more into the future could have its uses—or, at the very least, appeal to a certain kind of adventurer.
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)
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.
Quelques mots sur le cuistot
J'ai écrit plus de 50 articles ces dernières années, à retrouver ici, dont une bonne partie publiés dans des médias comme le Journal du Net (mes chroniques ici), le Huffington Post, L'Express, Les Échos.
Retrouvez ici mon podcast Parlons Futur (ou taper "Parlons Futur" dans votre appli de podcast favorite), vous y trouverez entre autres des interviews et des résumés de livres (j’ai notamment pu mener un entretien avec Jacques Attali).
Je suis CEO et co-fondateur de l'agence digitale KRDS, nous avons des bureaux dans 6 pays entre la France et l'Asie. Je suis basé à Singapour (mon Linkedin, mon Twitter), également membre du think tank NXU.
Merci, et bon weekend !
Thomas