Discover more from Datascience Learning Center
What is Kyutai?
Could this become the OpenAI of Europe?
With all the drama surrounding Sam Altman right now in November, 2023, I’d venture to guess that the most important person in AI is actually former Google CEO Eric Schmidt. From policy, to venture capital to other big bets.
It’s turning out that Aleph Alpha is more like the Anthropic of Europe and Mistral seems to be more on the open-source side of things. What will Kyutai evolve into? While I’m not exactly proud of how Eric Schmidt has turned out, he is throwing his weight and fortune around in some interesting ways.
Sovereign AI is still just a Dream
The French are an idealistic bunch, but are Billionaires idealists too? Billionaires Xavier Niel, Rodolphe Saadé and Eric Schmidt announced a new nonprofit artificial intelligence research lab in Paris, marking France’s latest push to develop sovereign AI technology.
Buried beneath the OpenAI hype and drama, in Europe something was occuring that was much more interesting.
If the U.S. is centralizing A.I. without real regulation under the control of BigTech, in Europe you have a counter culture, they are celebrating open-source more Hugging Face style. It has a french vibe.
Kyutai is a French AI research lab with a $330 million budget that will make everything open source
But look at the phrasing of their word choice:
OpenAI started out with idealistic foundations as well, and just look at how it turned out.
Kyutai, a non-profit laboratory, is focused on open research in AI and aims to develop large multimodal models that utilize text, sound, images and other data types, while also inventing new algorithms to enhance their capacities, reliability and efficiency, according to the release.
Sounds familiar eh! But what with OpenAI with European characteristics? If the U.S. has taken LLMs to the corporate cloud, moat and locked, what can the world do to counter that A.I. weaponization of greed?
A private, non-profit initiative
Kyutai was founded jointly by the iliad Group, the CMA CGM Group and Schmidt Futures.
To achieve this, Kyutai will utilize the computing power provided by Scaleway, a subsidiary of the iliad Group, which boasts the highest-performance computing power for AI applications in Europe, the release said.
It will work with PhD students, postdocs and researchers on research papers and open source projects. When Iliad originally unveiled this research lab, the firm said that Niel was committing €100 million to this project ($109 million at today’s exchange rate).
My prediction is this becomes the ‘DeepMind of Meta’, Meta also likes to pretend it’s really into supporting open-source. Watch Facebook fund this in the near future in 2024.
Since the UK has been relatively disappointing in AI startups in Generative A.I., France is looking half decent with some big attempts like Mistral, Hugging Face adn now Kyutai.
According to Xavier Niel, chairman and founder of the iliad Group, Europe has the potential to excel in the AI race, and creating an AI open research lab in Paris is crucial to leverage the expertise of French researchers, available computing power and the dynamic AI ecosystem.
Billionaires for Science
Three billionaires, Eric Schmidt, Xavier Niel, and Rodolphe Saadé, announced the launch of the lab.
The main reason Eric Schmidt is important, is his emphasis on AI in the development of Quantum computing and Science. He was important in the Biden Administration banning A.I. chips to China. Way more important than Sam Altman in the grand scope of things, sorry for the AI bros reading this.
What is their Mission?
“OUR MISSION IS TO
BUILD AND DEMOCRATIZE
ARTIFICIAL GENERAL INTELLIGENCE
THROUGH OPEN SCIENCE”
I have no idea what that really means, but three things are important here, that they name AGI and Open and Science.
Europe will try to donate its way to open-competition. TechCrunch says that this is just a starting point, as Kyutai is open to more donations. “What’s interesting with so many journalists in the room is that the project will potentially interest other investors,” Saadé said at a press conference after the announcement.
For Billionaires, they want to buy a better legacy. They want their wealth to matter and translate into the future in a sustainable way.
Kyutai Got the Chips
As Kyutai will work on foundational models, they will also need some compute power. The good news is that Scaleway, the cloud division of Iliad, recently acquired a thousand Nvidia H100 GPUs. These top-of-the-line GPUs are essential for inference and model training and will be available at cost for Kyutai.
Implications for the Future
What if we had foundational models geared for science? Not Elon Musk’s version of what means, but Europe’s? The sort of AI that could make AI research laboratories around AGI.
The creation of Kyutai is not only a milestone in France’s tech journey but also a testament to the growing importance of AI in various industries. With its strong scientific team and focus on open-source research, the lab is well-positioned to contribute to foundational models and projects in AI. The founders hope that the lab’s establishment will not be impeded by regulatory constraints, thus, facilitating unhindered technological progress in the realm of artificial intelligence.
Eric Schmidt is backing Future House, among many similar projects. The biggest impact of LLMs is likely going to be in science and innovation itself, in how R&D occurs.
Schmidt has an estimated net worth of $24.5 billion, according to the Bloomberg Billionaires Index.
If Google were a good company, which it is not, its Search Advertising revenue would mean something benevolent to the future. People like Eric Schmidt have leverage of how the internet got started and are the types of people who profiteered from Silicon Valley’s centralization around early winners like Google, who up until recently, was perhaps the dominant “pure-play AI” company of the world. But a lot has changed since that 2017 Google paper. All the researchers are gone, a lot of the talent has left soon after.
We need to rebuild what AI could be, and might become with a bunch of smaller bets on things like AI and Science.
This might be the legacy of Schmidt after all. Nobody is more responsible for the U.S. vs. China momentum in AI. For good or ill.
Credentials, Plans and Building
Kyutai has already started hiring for its core scientific team. Six men took the stage this morning to talk about their previous work and what they have in mind for the research lab — Patrick Perez, Edouard Grave, Hervé Jegou, Laurent Mazaré, Neil Zeghidour and Alexandre Defossez.
They previously worked for Meta’s AI research team FAIR, Google’s DeepMind division, Inria, etc.
The Facebook-Meta connection is actually thing:
Patrick Perez, who previously worked for Valeo, is going to be the director of the research lab.
Kyutai has also put together a team of scientific advisors who are well-known AI researchers — Yejin Choi, Yann LeCun and Bernhard Schölkopf. They will just check everyone’s work once or twice a year and give feedback.
Eric is nearly 70, his productive years are nearly behind him. He tried very hard in the last decade to make his mark.
He’s funneled some of that fortune into philanthropic efforts like Schmidt Futures, an initiative that funds science and technology entrepreneurs. In recent months, he’s emerged as an influential voice on AI policy in Washington. He certainly tried to be a mover and shaker in ways that even the media has missed.
A media that has tried to keep up with him.
Follow Kyutai on X here.
One of the reasons Kyutai thinks it can convince some researchers to join its lab is that researchers will be able to publish research papers.
Speaking at ai-Pulse, an event organised by Niel in Paris’s Station F startup campus on Friday, Schmidt said he backed Kyutai because he believes France can become an AI champion, as long as it can keep regulators in Brussels “under some level of control”.
So there’s also an element of lobbying here, to push America’s agenda of a lack of regulation on to Europe, to make France a bit less like Germany in this regard. The battle over A.I. regulation is deeper than it seems.
Support from the French Government
French President Emmanuel Macron and Digital Minister Jean-Noel Barrot have expressed their support for this initiative that seeks to prevent Europe from falling behind in AI.
The effort could lay the groundwork for an artificial intelligence industry in Europe. Although honestly Aleph Alpha and Mistral are doing that now.
The top UK Generative A.I. startup is likely Stability AI, that has been fraught with controversy too. It had so much trouble getting funding, it has to go with Intel in the end. Very bizarre.
Europe has been left behind more or less even as China has been hobbled by poor timing and geopolitical and economic downturns to keep up with the U.S. in the all important Generative A.I. front.
Rodriques, a biotechnology inventor who studied at the Massachusetts Institute of Technology, said Schmidt will fund Future House for its first five years. FutureHouse is a bit simliar. It’s a nonprofit that’s focused on building an artificial intelligence-powered assistant for the laboratory, with the lofty goal of overhauling the scientific research process. Read more.
Basically as OpenAI has gone commercial and even Anthropic has sold out to the Cloud in terms of Amazon and Google, more organizations are cloning the model of the old OpenAI, before it turned all corporate rogue under Sam Altman, under the watchful eyes of Satya Nadella with his chariots of Copilots. Which in a few years’ time might sound very lame, as lame as Cortana became in the last round of chatbot history.
Kyutai, FutureHouse and even X.AI seek AI at the intersection of science and human progress. There will be many others.
But Schmidt is a bit anti-European and pro-American. For example:
The EU AI Act
Schmidt — who is well-known for his criticism of EU regulators — pointed to the AI Act, which is currently being negotiated in Brussels, as an example of laws that might break innovation. But innovation does not occur at the service of a profiteering Cloud either with U.S. boardrooms and executives?
OpenAI, as the name still indicates, started as a nonprofit. But things changed drastically after Sam Altman started working full time on OpenAI in 2019. Perhaps OpenAI’s board relegalized the error years too late. Now it’s more like the DeepMind of Microsoft, or a wing of Microsoft Research focused on commercialization and poaching Google talent, basically serving Azure.
It’s near the end of 2023 that you realize how political AI has become. How fragmented and so much about power. The power of nations, the egos of billionaires, the dogmas of corporations.
Does Kyutai think it has enough H100s though? Iliad had acquired 1,000 Nvidia H100 GPUs that would be available to Kyutai to power the research of the new outfit. Compute, funding, competition - all in the name of ideals and science. What could possibly go wrong?