Transcription
I said earlier AI race and you didn't push back. So I'm curious. Do you buy that as a metaphor? You know, is there going to be a winner or is this something where Europe can build value, build growth without coming in first place ever?
[snorts] >> Well, you need to be in first place in certain domains which is what we have been doing in certain uh areas of intelligence like audio processing documents uh doing symbolic reasoning as well. We have a model doing a uh symbolic mathematics and mean uh so there are areas I mean the AI intelligence surface is very large uh you can go in multiple directions you can make them better at physics you can make that make them better at life science at coding so you get to pick a little bit where you want to be very strong uh so we think that uh we we believe that in Europe AI can be very strong in particular when combined with the manufacturing world because that's a place where Europe has been historically strong and that's what we're doing with our our customers not your question will there be a unique winner I think it's useful to look at uh the economics of AI if everything go well and so if effectively this becomes a powerful enough technology to drive let's say only 10% of human labor today um if you look at um the the wages the total number the total amount of wages uh in uh in Europe this is around 9 trillion uh if you assume that in five years or even 10 years uh you 10% % of these uh is actually AI uh because really AI cost is comparable to human labor in many domains.
Um well that's around one trillion. Uh if uh all of these flow back to say the US uh you're basically increasing very very significantly uh the commercial imbalance um to a point where this creates enormous instabilities. Uh at some point the the the commercial balance of the world is also the reason why we live at peace. Uh and so this is not acceptable uh for any region. uh I'm thinking of Europe but I'm also thinking of all Asian countries India uh Africa to fully depend on uh a single provider and not to create some of the IP itself not to do R&D itself I think that's generally the case the reason why this is the case is that this is a big market you're comparing AI should be comparable in 10 years to let's say I'm I'm kind of this this is orders of magnitude but it's 10% of human wages and that's enormous and so you don't have uh two players uh that are providing the entire world with such a large technology. It is very comparable I think in that respect to energy. Uh you don't have a single energy provider in the world uh that is monopolistic. The reason for that is that this is too big a market and the second reason is that you also need to have business continuity. uh you want to make sure that your electricity providers when you're in Europe are actually European because they don't want to be cut off and we've seen that even like at the at the molecule level this was really problematic uh to depend too much on others. So AI should be compared to uh electricity and to to some extent should be also compared to molecules.
Uh so that means we will continue every country will continue to import technology from other places in areas where the models are effectively stronger. So we export to U to the US in manufacturing for instance. Uh and uh it will produce its own technology uh because it needs to uh every country needs to make sure that whatever happens the AI can continue to run and it will uh and it will export technology. So you import you export you produce. It's exactly like energy and because of the size of the market it's pretty clear that it will fragment. If it doesn't fragment, you're running into an extremely unstable world.
>> When you tour Europe, speaking to politicians about this, giving your view, are there any who really get it?
>> Uh, it depends on the people, I would say. Uh, I think uh all of the countries we work with uh are getting it. Uh if you talk to I mean France has been where we started to so they got it pretty quickly. Uh but we're also working with Luxmbborg, with Greece, we're working with Sweden. uh we're working with Spain and so those countries uh are getting the fact that AI is a massive vector of power the same way oil was a massive vector of power and will define uh the geopolitics of the world. If you look at the the eastern countries as well, uh they see AI as a major component in defense systems and they they need those defense systems and so that's uh the realization that AI is a sovereign technology for economic purpose but also for strategic purpose is driving a sense of urgency. Now of course as you know uh the European Union is not really meant to do industrial policy. uh it's uh uh it's meant to try and drive consensus and typically consensus makes you a market doesn't make you a producer. Uh and I think they're realizing they need to change things. Uh but the institutions makes it pretty slow. So what we see is that there's actually more realization at the country and state level than at the European Union level. European Union actually re realizes that it needs to do things but structurally it cannot go very fast.
Uh so our our approach there has been to really partner with countries show what AI can bring to those countries because really if you if you want the thing to be acceptable uh you need the technology to be deployed to citizens they need to think that suddenly their public services is improving because of artificial intelligence. If you only go and tell them look it's going to take away your job it's not acceptable. Uh so the the approach we've taken with countries is first equip them in a way that is sovereign so that they actually own the keys uh can deploy it in defense systems uh can maintain the continuity of the service whatever happens and then help them in figuring out the public services that could be equipped uh so it can be finding a job uh it can be understanding the law uh it can be uh dealing with the social security services uh it can be dealing with your taxes so all of these things are kind of cumbersome for for citizens today. Turns out if you use artificial intelligence, you can you can make them uh faster uh you can make them more more agreeable and more delightful for for citizens. And so that shows them that effectively it can have an impact. The the second thing we do is because of you know like in in um in Europe a big and transversal problem is the fact that uh we're growing old and uh that uh the states are actually using a lot of civil servants uh and so with AI you can actually not replace but at least uh make them more productive so that you can deal with the the fact that people are just retiring.