Here's a look back at last Thursday's Bsmart TV show...

• Beware of "Open Source washing"...

Just as everyone wants to be "Green", with a lot of "greenwashing" in the process, there is now a lot of "Open Source washing". We have to be careful! [...] Many of the so-called Open Source LLMs are at best 'Open Science', meaning that they limit the possibilities of use under certain conditions.

For example, you have the right to use the software, but you do not have the right to use it commercially. And above all, they deliver the model, they deliver the training methods, but they don't deliver the data sets ....

Let's be very clear: a truly Open Source AI is one to which you have access: 

- the models
- the training methods
- to the data sets.

This is what the OpenLLM France community is doing.
 

At LINAGORA, we are working with Open Source to define what we can call Open Source AI.

• When it comes to AI too, digital biodiversity is essential - we need to 'change the game'!

I'm convinced that Mistral is going to be very successful, and I wish them every success, and so much the better for having French champions. But my deepest conviction is that Mistral will not be the champion of digital sovereignty in the field of French and European AI, and that's why I think we need to break the codes... We need to completely change the game, and the only model that will enable us to stand up to the American giants is truly Open Source models, and that's why we need to invent new ways ...

• Yes, it's possible, it's not inevitable!

We need to make things differently, so we don 't have to rely on just one player, and that's where cooperation between the public and private sectors is essential, especially when it comes to computing resources. When you use Open Source, you have access to computing resources that are simply gigantic.

In France, we have an organisation called GENCI, which manages the famous Jean Zay supercluster, and which has just purchased additional clusters. So YES, the reality is that a third way is possible and that we need to work in that direction and , above all, not get discouraged. There's no reason why we can't succeed.

Once again, a big THANK YOU to the teams...