How a missing gene in our DNA affected the Battle of Trafalgar and gave birth to the Mafia
Tech to regrow new teeth; See what that machine uses AI and laser for & more
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, plus d'infos sur moi en bas d'email.
Voici donc ma dernière sélection ! Avec la réponse à l’objet de la newsletter, plutôt sur le thème “Parlons passé” pour une fois, en fin d’email :)
L’apéro
Threads, the Twitter-killer app launched by Meta/Facebook is about to reach 100 million users just a few days after its launch, for context (source)
It took ChatGPT 2 months, TikTok: 9 months, Instagram: 2,5 years, Facebook: 4,5 years & Twitter: 5 years
In one interesting experiment, GPT-4 Outperformed Humans in Pitch Deck Effectiveness Among Investors and Business Owners
Demis Hassabis, DeepMind’s cofounder and CEO, says his engineers are using techniques from AlphaGo to make an AI system dubbed Gemini that will be more capable than that behind OpenAI’s ChatGPT. (Wired)
Hassabis says his team will combine Large Language Model technology with techniques used in AlphaGo, aiming to give the system new capabilities such as planning or the ability to solve problems.
After beginning to plateau in May, ChatGPT's website traffic dropped around 10% from May to June. (source)
Meta has unveiled a "breakthrough" new text-to-speech AI that can edit existing audio, speak in 6 languages, and replicate your loved ones' voices. (source)
all you have to do in order to replicate someone's voice is feed the program an audio clip as short as 2 seconds long.
Meta argued that the tech could "allow visually impaired people to hear written messages from friends in their voices."
but it could give way to phishing scams, misinformation, and even an audio version of deepfaked porn.
Thankfully, Meta is more than aware of this, and is opting to keep the model and its underlying code closed source for the time being.
Bild, the German tabloid, is expected to replace over a hundred human editorial jobs with artificial intelligence (The Guardian)
The tabloid will "unfortunately be parting ways with colleagues who have tasks that in the digital world are performed by AI and/or automated processes,"
Steve Jobs was already thinking of a VR headset back in 2007 (listen to him in this 34-sec video)
Hum... A significant proportion of people paid to train AI models may be themselves outsourcing that work to AI, a new study has found. (MIT Tech Review)
Using AI-generated data to train AI could introduce further errors into already error-prone models.
Medication that may allow people to grow new teeth moves towards clinical trials in Japan (source)
See this impressive 35-sec video of that machine in action: a vehicle that uses AI and laser to destroy weeds (and not crops) in fields, at sub-millimeter accuracy, developed by CarbonRobotics
2-min video of a a robot assembling an IKEA chair
Robot is not autonomous, but is controlled by a human using haptic controllers (seen in the video)
Video is sped up, but still, it took less than 8 minutes to assemble the chair (original video)
If such nimble robots become affordable, that means in the future human teleoperators could help fix all sorts of issues at home remotely
And of course, once enough data of interventions is in, at some points AI can be trained to operate these robots. It will take time though.
There are 3 million software engineers in Europe in 14 skill areas, according to Sequoia Capital
Top 5: 10% in London, 2.74% in Paris, 1.95% in Madrid, 1.76% in Amsterdam, 1.75% in Berlin
Research firm Rystad Energy now predicts that annual installations of battery energy storage systems will grow tenfold between 2022 and 2030 (equivalent to 33% growth rate).
In May, the EU’s combined wind and solar power generation surpassed fossil fuels for the first time. (source)
Découvrez dans cette vidéo d'une minute comment l'IA de Best Western, développée par mon entreprise YeldaAI, répond aux clients au téléphone en langage naturel (et non avec les options "tapez 1, tapez 2, etc.")
Tech analyst Ben Evans on Meta's and Apple's approach to VR/AR
Meta, today, has roughly the right price and is working forward to the right device
Apple has started with the right device and will work back to the right price.
New McKinsey's report: about 75% of the value that generative AI use cases could deliver falls across 4 areas: Customer operations, marketing and sales, software engineering, and R&D
India now has the highest data usage per user on earth, and the lowest prices. (The Economist)
Someone built a convincing-sounding voice powered by OpenAI's GPT-4 that can waste and frustrate telemarketers and scammers by happily engaging in long and pointless conversation. (WSJ)
Very cool videos of the huge sphere built in Las Vegas: here, here and there
An entertainment arena with a capacity for 18,000 people
With more than 100,000 LED screens on the outside, making it unique.
The MSG Sphere in Las Vegas is the world's largest spherical structure
The Sphere is 112 m high and 157 m wide at its broadest point.
Massive underground deposit of high-grade phosphate rock in Norway, pitched as the world’s largest, is big enough to satisfy world demand for fertilisers, solar panels and electric car batteries over the next 100 years, according to the company exploiting the resource. (source)
Road transport is responsible for 50% of oil consumption worldwide, but if we were to electrify 100% of road transportation at current levels, we would only need 22% as much energy... because electric cars are so much more efficient at converting energy to motion, 4.5 times on average actually it seems!
Mon intervention en vidéo dans le cadre du think tank NXU sur les différentes “factions” de l’IA
Mon dernier article dans le Journal du Net : L'IA va-t-elle provoquer un chômage de masse ? Creuser les inégalités ? Renforcer les GAFA ?
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
Pourriez-vous me dire en retour d’email si vous seriez plus enclin à partager cette newsletter si elle était traduite entièrement en français ?
À table !
Self driving cars are here at last: In some San Francisco neighborhoods, at certain hours of the night, it seems as if one in 10 cars on the road has no driver behind the wheel. (MIT Tech Review)
These are not experimental test vehicles, and this is not a drill. Many of San Francisco’s ghostly driverless cars are commercial robotaxis, directly competing with taxis, Uber and Lyft, and public transit.
I’ve come to believe that most people, including many powerful decision makers, are not aware of how quickly this industry is advancing, or how severe the near-term labor and transportation impacts could be.
While not perfect, my most recent Cruise ride, in April, was sufficiently close to the experience of riding with a responsible human driver that I momentarily forgot I was in a robotaxi. The mere fact that these vehicles are programmed to follow traffic laws and the speed limit automatically makes them feel like safer drivers than a large percentage of humans on the road.
In international tests pupils in Vietnam has been doing much better than youngsters from much richer countries for a decade. (The Economist)
Vietnam’s children spend less time in lessons than Indian ones, even when you count homework and other cramming. They also put up with larger classes.
The difference is that Vietnam’s teachers are better prepared, more experienced and more likely to be held accountable if their pupils flunk.
Vietnam’s teachers do their job well because they are well-managed.
They receive frequent training and are given the freedom to make classes more engaging.
To tackle regional inequality, those posted to remote areas are paid more.
Most important, teacher assessment is based on the performance of their students. Those whose pupils do well are rewarded through presitigious “teacher excellence” titles.
Modern timber(en bois) buildings can be cheap, green, fireproof, and faster and less noisy to build than concrete(The Economist)
a 100-metre-tall wooded skyscraper will soon be built in Switzerland
a Swedish firm unveiled plans to build Stockholm Wood City, what could be the world’s biggest wooden city.
Construction will begin in 2025.
When complete, 10 years later, it will contain 2,000 homes and 7,000 offices, along with restaurants and shops.
By using wood the company hopes to reduce the project’s carbon footprint by up to 40%, compared with building in concrete and steel
engineered timber is a composite in which layers of wood are laminated together in specific ways. The wood grains in each layer are aligned to provide individual components of the building, such as floors, walls, cross braces and beams, with extremely high levels of strength.
The use of prefabricated sections cuts down on the delivery of raw materials and allows construction to proceed more quickly.
Another advantage is that construction will not be as noisy as it would be if the town were built from concrete and bricks
Researchers are coming to believe that engineered timber is, by its nature, extremely fire resistant.
When it comes to LLMs, China trails years, not months, behind its international competitors (Foreign affairs)
The Chinese semiconductor industry can only produce chips several generations behind the latest cutting-edge ones
many Chinese startups are now opting to base their operations overseas and sell to an international market instead of selling primarily to the Chinese market
Chinese AI companies are also subject to the country’s unusually detailed and demanding regulatory regime for AI.
In a private conversation with one of us (Xiao), one Chinese CEO quipped that China’s LLMs are not even allowed to count to 10, as that would include the numbers eight and nine—a reference to the state’s sensitivity about the number 89 and any discussion of the 1989 Tiananmen Square protests
Some of the rules may prove quite onerous. The January regulations, for instance, oblige providers to “dispel rumors” spread using content generated by their products, meaning that companies are on the hook if their AI tools produce information or opinions that go against the Chinese Communist Party line. The April draft would go further still, forcing LLM developers to verify the truth and accuracy not just of what the AI programs produce but also of the material used to train the programs in the first place
European Union's AI Act : 5 major implications (MIT Tech Review)
Ban on emotion-recognition AI. The European Parliament’s draft text bans the use of AI that attempts to recognize people’s emotions in policing, schools, and workplaces. Makers of emotion-recognition software claim that AI is able to determine when a student is not understanding certain material, or when a driver of a car might be falling asleep. The use of AI to conduct facial detection and analysis has been criticized for inaccuracy and bias, but it has not been banned in the draft text from the other two institutions, suggesting there’s a political fight to come.
Ban on real-time biometrics and predictive policing in public spaces. This will be a major legislative battle, because the various EU bodies will have to sort out whether, and how, the ban is enforced in law. Policing groups are not in favor of a ban on real-time biometric technologies, which they say are necessary for modern policing. Some countries, like France, are actually planning to increase their use of facial recognition.
Ban on social scoring. Social scoring by public agencies, or the practice of using data about people's social behavior to make generalizations and profiles, would be outlawed. That said, the outlook on social scoring, commonly associated with China and other authoritarian governments, isn’t really as simple as it may seem. The practice of using social behavior data to evaluate people is common in doling out mortgages and setting insurance rates, as well as in hiring and advertising.
New restrictions for gen AI. This draft is the first to propose ways to regulate generative AI, and ban the use of any copyrighted material in the training set of large language models like OpenAI’s GPT-4. OpenAI has already come under the scrutiny of European lawmakers for concerns about data privacy and copyright. The draft bill also requires that AI generated content be labeled as such.
That said, the European Parliament now has to sell its policy to the European Commission and individual countries, which are likely to face lobbying pressure from the tech industry.New restrictions on recommendation algorithms on social media. The new draft assigns recommender systems to a “high risk” category, which is an escalation from the other proposed bills. This means that if it passes, recommender systems on social media platforms will be subject to much more scrutiny about how they work, and tech companies could be more liable for the impact of user-generated content
What happens next? This does not mean the EU is going to adopt these policies outright. Next, members of the European Parliament will have to thrash out details with the Council of the European Union and the EU’s executive arm, the European Commission, before the draft rules become law. The final legislation will be a compromise between three different drafts from the three institutions. European lawmakers are aiming to get the AI Act in final shape by December, and the regulation should be in force by 2026. (Politico)
Very strong negative reaction from 150+ European executives from the likes of Renault, Heineken, Airbus, and Siemens. (The Verge)
In a one-page open letter regarding the recent EU AI Act, they say that it jeopardises competitiveness and technological sovereignty; that it suppresses opportunity in Europe; and it threatens the spirit of innovation that’s bubbling right now.
The fundamental reason: companies developing foundation models will have disproportionate compliance costs and liabilities.
"Such regulation could lead to highly innovative companies moving their activities abroad, investors withdrawing their capital from the development of European Foundation Models and European AI in general. The result would be a critical productivity gap between the two sides of the Atlantic."
Parlons passé : How a missing gene in our DNA affected the outcome of the Battle of Trafalgar and gave birth to the Mafia. (source)
Sometime between 60 and 40 million years ago, a particular gene was knocked out by a mutation in our tree-dwelling ancestors, and so is a defect shared by all humans. The loss of this gene means that humans cannot create vitamin C for ourselves, and so we must get all we need from our diet: else we begin to suffer from the deficiency disease known as scurvy (scorbut).
In fact, because most other animals are able to produce vitamin C, scurvy is almost a uniquely human disease.
It has been estimated that between 1500 and 1800 scurvy killed over two million sailors – accounting for more deaths at sea than storms and shipwrecks, naval battles and all other diseases put together.
The defeat of scurvy was a major contribution to the success of the Royal Navy during the Napoleonic wars, including the effective blockading of French and Spanish fleets in their ports and then victory at the Battle of Trafalgar. As naval historian Christopher Lloyd put it, ‘Of all the means which defeated Napoleon, lemon juice and the carronade gun [a naval cannon] were the two most important.’
The adoption of citrus juice rations by the Royal Navy, however, created a huge surge in demand for lemons, and in 1803 Lord Nelson turned the island of Sicily into a giant lemon factory. The problem was that the high value of lemons, combined with the fact that at the time the island was poorly governed and that the fruit are very easy to steal from orchards, meant the growers were forced to hire private muscle to provide security. This evolved into a system of extortion and protection rackets - and by the 1870s the organisation we would recognise as the modern Mafia had emerged on Sicily.
An amazing debate opposed 2 of the 3 godfathers of deep learning, here is what happened:
Proposition debated "AI research and development poses an existential threat." (video of the debate that lasted 1h45min)
Pro:
Yoshua Bengio: Full Professor at Université de Montréal, and the Founder and Scientific Director of Mila – Quebec AI Institute (and one of the 3 godfathers of deep learning)
Max Tegmark: Professor doing AI and physics research at MIT
They were very much aligned, their argument in a nutshell:
AI will match and likely go beyond human intelligence
We don't know how to control superhuman intelligence, we may not be able to, a super AI will likely develop sub goals not aligned with humans, which can lead to devastating, existential consequences
3 main issues: malicious use of powerful AI by genocidal humans ; AI escaping our control ; and AI outcompeting us in the economy (and to which we cede more and more control in our world until the point we're too dependent on it)
Con:
Yann LeCun: VP & Chief AI Scientist at Meta and Silver Professor at NYU (and one of the 3 godfathers of deep learning )
Melanie Mitchell: Professor at the Santa Fe Institute
They were not exactly aligned (on the question of whether we'll get superintelligence in the first place)
Yann LeCun's argument:
we'll get super AI for sure, we'll build it incrementally and ensure it's safe, one step at a time, it's an engineering problem that we can fix
He said at some point "if we can't control it, we won't build it, right?" ... the room burst into laugh
Melanie Mitchell's argument:
super AI is likely very far away and not sure it's possible, no indication we'll get there, so the existential risk is very low, such concerns suck the oxygen out of the room at the expense of other more pressing risks around bias, discrimination, disinformation, etc.
A super AI can't be super smart and yet so dumb to kill us all, will likely understand not only the letter but also the spirit of our laws
Not there at the debate, but to note, the 3rd godfathers of deep learning, Geoffrey Hinton would have been in favor of the proposition
The PRO/CON split among the audience was 67/33 before the debate and 63/37 after. So the CON side won the debate by 4%, a modest but notable gain. Though most people still thinks there's existential threat
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.
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.
Je suis CEO et co-fondateur de l'agence digitale KRDS, nous avons des bureaux dans 6 pays entre la France et l'Asie, ainsi que de Yelda, “A voice assistant to answer every call” (notre itw sur BFM Business “À la mairie de Plaisir, c'est l'intelligence artificielle qui répond au téléphone”), une des Future 40 startups de Station F.
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 basé à Singapour (mon Linkedin, mon Twitter), également membre du think tank NXU.
Merci, et à bientôt !
Thomas