The accelerated race for large-scale artificial intelligence developments has found Europe on the wrong foot and it is feared that it will be left behind, but in the face of giants like Microsoft, Google or Meta, the old continent still has hope in special AI, according to experts.
Microsoft has incorporated a new version of chatGPT, a powerful model developed by the company OpenAI in Bing, while Google has opened its rival software Bard to the public, and Meta is again focusing on AI.
As searched by guns, Europe is the first to “cut off” the train of artificial intelligence, laments Francis Soulié-Fogelman, pioneer of the sector in France.
“We are not going to create another GPT chat. We do not have those companies with great resources. We have lost a great public,” he explains.
But “with companies, it’s a little different. We have big tech companies and talents” that “big techs” like Google, Apple or Facebook “steal from us. It’s hard to find chips outside of us,” he says. scientific advisor Hub France IA, which brings together hundreds of companies in the region.
Paul-Francois Fournier, director of innovation at Bpifrancos, makes a similar analysis. “In the United States it has achieved a predominant position, but we will have applications for industries where Europe has strong positions, such as aeronautics or automobiles,” he explains.
It can still “build the champions of the European world,” the former AI director at Facebook parent Meta, Antoine Bordes, who joined the German military artificial intelligence company Helsingin last week, said on Thursday.
As an example of the future, the ambitious LightOn launched on Friday a custom ChatGPT for companies, and the company Illuin Technology, a language expert, proposes AI analysis of conversations with patients at hospitals.
But in this region, the European giants will also offer: NVidia, the leader of AI super-microchips, announced this week the launch of a “model machine” to allow each team to create its own AI. IBM promises companies “to order data” for specialized AIs.
Continue reading the story
– European AI Regulation –
The European Union has provided 10,000 million euros (about 10,780 million dollars) for AI between 2014 and 2027 and its 2021 AI Plan for public and private funds has provided 20,000 million euros (USD 21,560 million) per year over the decade.
Despite this, Europe is a hundred billion far from the Americans and the Chinese and the Americans.
“Every year OpenAI, Microsoft or Meta invest in artificial intelligence (…) more than the whole of Europe invests in research”, lamented this week the former president of the French public group, Maurice Lévy.
The European Court of Auditors launched an audit on Monday to determine whether the European Commission’s investments would allow Europe to achieve world-leading status.
But the EU has a powerful weapon: the future regulation on Artificial Intelligence, the world’s first organization in this field, to protect citizens, their privacy and democracy.
The European Parliament could vote on the first version in April to start applying it in 2025.
The principle is to create a certificate for all AIs as “high risk”, a category that is still being spread – the cause of generative AIs has not yet been defined -, without which they will be banned in Europe.
This shield broke the appetite of the Chinese and Americans, but also in European innovation.
“The commission has estimated the cost of the certification at 300,000 European coins,” warns Anissa Kemiche, who represents the French professional confederation Numeum Brussels.
“It will be necessary to go to the market with a seal, the opposite approach to the Americans… Google and Microsoft will arrive with thousands of advocates, but how will they do the small operations?”, concerns Françoise Soulie-Fogelman.
“These regulations can help create a complete local ecosystem, with a supply of local AI microchips,” says the CEO of Intel France, Erwan Montaux, whose group will open a 20 billion euro semiconductor megafactory in Germany in 2024.
leb/pel/aoc/mar
