Sovereign AI as a Societal Reflection

There is a real tension around AI development in the US; the AI "arms race" as some call it, and the race to the top (or bottom, depending on how you look at it).

An article in Tech Policy Press (link in comments) lifted up the three models of AI development most prevalent today in the US, China, and Switzerland.

🇺🇸 US: private frontier labs racing to scale, funded by investors who incentivize constant one-up-ing to justify another investment round. The incentive is 💰 to keep raising the ceiling, and put a new model in front of you constantly.

🇨🇳 China: giving away smaller, cheaper, open-weight models. The strategy is models that reach more people. Wide adoption now. Possibly national influence later.

🇨🇭Switzerland: a model called Apertus, built by public universities and supercomputing centers, trained only on licensed data, run on hydropower. To advance the public good.

Three different models of AI development with no collaboration.

But sovereignty over cooperation isn't inevitable... it's a series of choices being made right now by people who benefit from things staying the same.