Guillaume Grallet, Tech & Sciences editor-in-chief at “Le Point,” welcomes Aidan Gomez, cofounder of Cohere, a Canadian artificial intelligence company that develops language models and tools mainly intended for businesses. The entrepreneur today rejects the end-of-the-world scenario oversold by Silicon Valley.
§ #IA #intelligenceartificielle #Cohere #AidanGomez #startup #tech
Follow us on:
Youtube : / lepoint
Facebook : / lepoint.fr
Twitter : / lepoint
Instagram : / lepointfr
Tik Tok : / lepointfr
LinkedIn : / posts
Bluesky : https://bsky.app/profile/lepoint.fr
Threads : https:/...
What Aidan Gomez said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Aidan Gomez, CEO and cofounder of Cohere, argued that democracy must lead in AI and that capabilities should be disseminated beyond one democracy to avoid a single point of failure. He said the AI race risks collapsing to a one-player race and urged middle powers like Canada, Germany, and France to pool resources and build a diverse supply chain, rather than each country building everything. On learning, he advised doing the hardest things possible to learn how to learn. He predicted AI will increasingly crack long-open math theorems but humans will choose which problems to work on. He dismissed fears of a 'cognitive apocalypse' and said world models will be part of LLMs, not a separate track. He identified continual learning as the next step in AI. He recommended books by Penrose and Stephenson as formative.
Key takeaways
Middle powers like Canada, Germany, and France must pool resources to build a diverse AI supply chain, not isolationist self-sufficiency.
AI is already cracking theorems open for 100-200 years, but humans will still choose which problems to work on.
World models will be a part of LLMs, not an entirely independent track.
Numbers and commitments
Figure
What it refers to
Type
At
100 years
Age of theorems being cracked by AI in mathematics
6:53Can we imagine in maybe 30 years or 50 years the AI being able to do mathematics alone and leaving the human being not able to do some progress, not able to think by itself?
8:34The fact that can have a representation of the world around them, do you think that's kind of world creates a different AI?
I think the most important thing that I'm going to tell my children is that you go to school to learn how to learn. You don't, it doesn't matter that much what you go to school for, what you learn, so long as it's difficult and it teaches you how to learn to do difficult things.
So Alan, thank you so much for being here. So you co-created Cohere. To what extent are the values of Cohere different from the industry? Why is it important for you to have an AI different from the one being developed today by the US or by the Chinese? And the question also about democracies, you told me they are important, that there are other democracies that are on apart from the US.
Yeah. Well, I think that it's very important that democracy in general continues to lead on the global stage and so we're going through a period of large-scale geopolitical change. And that can be a good thing, right? Like it could change for the better. We could see that democracies are actually stronger and the democratic alliance is stronger and more influential on the global stage on the other side of that change or it could get worse. And so I think it's in the interest of everyone who believes in democracy to ensure that it's not just one democracy that is doing all the work and developing all the capabilities because that presents a single point of failure. If that democracy fails, the whole project fails. So right now I think the challenge for Canada, for Germany, for France and everyone who still believes in democracy and this way of life is to ensure that capabilities get disseminated beyond just one democracy and that we are all capable and we all have something to contribute to the global stage. In AI, we run the risk of it collapsing to a one-player race. And so we really need to ensure that we are one of those players. We are one of the few that knows how to build this technology that can shape it into the future and can make sure that it protects the values and the interests of our people and our way of life.
Mhm. This is true for France, right? France should have also more important digital champions. And maybe very quickly, how could France go into that goal? Of course, France, you know, one thing that I would say is that sovereignty is not isolation and so it's not the case that what I mean is every single country needs to build everything, the entire AI supply chain, we all have to do chips, we all have to do models, we all have to do applications, that would be a impossible and b colossally inefficient and so that's not what I'm proposing. What I do think is that there needs to be a diverse supply chain. You need to have multiple providers and countries need to come together, middle powers need to come together to cultivate, pool resources and develop and invest in these capabilities. That way no one holds too much power over you. There isn't this single point of failure. Instead, if you have two people supplying you something, if one goes down or one disappears, you can move over to the other. That's what resilience and sovereignty comes from is that really there's no one else who has that much power over you. And so you have optionality and you can make your own decisions without the influence of anyone else. So that's what needs to be protected. That's what needs to be pushed forward. It's not that every country needs to do everything.
Very good. In these times of acceleration, there is one question everyone is asking too. What should we learn when we are 15, 20, 50 years old? And you had a unique childhood, right? In the forest with not good access to internet. Did it help? And to what extent, what should we teach our kids? What should make them read? For example,
I think the most important thing that I'm going to tell my children is that you go to school to learn how to learn. You don't, it doesn't matter that much what you go to school for, what you learn, so long as it's difficult and it teaches you how to learn to do difficult things. So, my advice to my brother and, you know, my sister-in-law has always been do the hardest thing you possibly can because regardless of if you're going to do that for the rest of your life or not, it will teach you how to do incredibly hard things and that skill can be applied anywhere and then you can do whatever you want, whatever the world needs from you.
I think I pursued something pretty hard like AI and computer science is a pretty technical field but yeah I guess like mathematics has always been the thing that's fascinated me. I've always thought it's been beautiful, creative and something that is very, very challenging.
Gomez, A. (2026, September 16). Aidan Gomez, cofounder of Cohere: “The AI race could come down to a single participant” [Interview transcript]. Le Point. CEOInterviews.AI. https://ceointerviews.ai/interview/1438944/
MLA
Aidan Gomez. "Aidan Gomez, cofounder of Cohere: “The AI race could come down to a single participant”." Le Point, 16 Sep. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/1438944/.
BibTeX
@misc{gomez2026_1438944,
author = {Aidan Gomez},
title = {Aidan Gomez, cofounder of Cohere: “The AI race could come down to a single participant”},
howpublished = {Interview transcript, Le Point. CEOInterviews.AI},
year = {2026},
month = {sep},
url = {https://ceointerviews.ai/interview/1438944/},
note = {Speaker-attributed transcript with timestamps}
}
