Macron & DeepMind CEO Demis Hassabis Speak at AI Summit in London | AI1G

Thank you very much. Well, I think you've touched on the competition in the AI space, and I think one of the key reasons for that as well is just the pace of innovation when it comes to AI. So perhaps Demis we could come to you and just get your views on how you respond to that pace and touching on some of the points just raised, what we need to be doing to not only keep up but to get ahead.

Thank you very much Arthur. If I could come to you on a similar point again, addressing this idea that the pace of innovation is something that is both a challenge and an opportunity and how you at Mistral AI think about that and respond to that.

Well, one of the premises of starting Mistral AI has been to commit to open science and bring open source models to the world. And it's interesting to note that we actually started the company as a French and British company as we incorporated the two companies almost at the same time. We had a workforce here very quickly. And to Demis's points, we need as countries to have a seat at the table to innovate. If you don't innovate, you're down to regulating and then it's a little less interesting and you have less of a say in how we use the technology. And so our approach there is to leverage the power of the two countries when it comes to training people and to have great scientists. And by committing to the open technology and building a business on top, building an enterprise business on top, we've been able to accelerate and change a bit the way the field was thinking about this issue of having enough transparency around the models, having enough accessibility to change them. And what we see is that with the models that we've released starting in 2023, the academic world has been able to use them quite extensively. Actually, we never had as many citations as with Mistral 7B, which is the first model we released. And the reason for that is that it's sufficiently small for scientists working in university to work on them and to do studies. And so we found that by committing to open source and by doing it more than what the US can do—the US is always a bit struggling whether they should do it or not because they have this competition with China, which we have a little less—by committing to this open technology, we've been able to spread science and to have a lot of new scientists work on that technology, and sometimes they came to us and started to work with us.

Thank you very much. I think the open source debate is going to continue and be an important policy area that we're going to need to balance. I think we've talked about the need for science and really allowing science to innovate and talking about how AI itself can help with that. I think the next question is how do we get some of these great breakthroughs to market? So, how do we have science for growth? Mr. President, perhaps you could say something about how France is approaching some of these challenges about translating these great breakthroughs from science from the lab and getting them to market.