Why Open Source AI Is Better for Quebec

There’s a fundamental difference between using proprietary AI (ChatGPT, Gemini) and open source AI. And for Quebec, open source is clearly the right choice. Here’s why.

Proprietary = you control nothing

When you use ChatGPT, you accept OpenAI’s terms. Your data? It goes to train OpenAI’s models. Your sensitive questions? They pass through American servers (and are subject to surveillance). OpenAI changes its prices whenever it wants. It changes its policies whenever it wants. You have zero power.

It’s like renting your house to someone who can change the rent tomorrow morning, and who wants to see everything you do inside.

Open source is the opposite. You download the code, you control it. You run it on your server. Your data stays with you. Nobody else sees it. And if you want to improve the AI, you can.

Sovereignty = not depending on the Americans

Quebec has a, let’s say, complicated history with cultural and economic domination. We’re among the best at resisting that. But with proprietary AI, we’re voluntarily putting ourselves under American dependency.

Open source means we can build our own AI. A Quebec AI, trained on Quebec data, that understands French, Quebec culture, our needs. Not a universal AI that puts everyone in the same mold.

Cost = much cheaper

OpenAI charges per request. Google charges per request. It adds up. For an SME or a nonprofit, it can get expensive fast. With open source, you have an initial setup cost, then it’s free. Or close to it.

That’s not nothing. It’s the difference between a startup that can use AI and one that can’t afford it.

But open source has challenges

Let’s be honest: the best open source models (like Meta’s Llama, or Mixtral) aren’t as good as ChatGPT4. They’re constantly improving, but for very specialized tasks, proprietary still wins.

And deploying open source AI requires technical expertise. You need someone who knows what they’re doing. It’s not a “click and go” like ChatGPT.

But it’s surmountable. And it’s an investment worth making.

What we should do

Invest in open source. Universities, governments, nonprofits — use open source. Train specifically Quebec models. Contribute to existing open source projects (Meta encourages the community to improve Llama). Create an ecosystem where open source AI is the norm, not the exception.

Yeah, it’s more work than paying OpenAI. But it’s a long-term investment in independence. And it benefits everyone.

That’s the conviction of Laeka Research: open source AI is better for our communities. To learn more about technology and independence in Quebec, visit Sherpa, our AI education platform in Quebec French.