It remains unclear to me if anyone funding new models will actually benefit economically from having funded these models.
I think is possible that these models can *culturally*(/memetically) benefit the creator via e.g. being trained to propagandise for the values of the originator, but it is unclear to me if this hypothesis has even been properly tested.
However, even absent benefit from creating new models, there are still good reasons to support the economic changes that would allow such models to be created in Europe:
We need a more unified, EU-wide, approach to investment both private and public; we need to upgrade the power grids even just because they're old and a poor fit for existing supplies and demands let alone forecast new demands and new supplies; and we need local chip factories to protect against supply chain problems in the event of conflict involving existing suppliers.
I expect the AI-training investment bubble to burst before the EU gets any of those things, those things are good regardless.
I think is possible that these models can *culturally*(/memetically) benefit the creator via e.g. being trained to propagandise for the values of the originator, but it is unclear to me if this hypothesis has even been properly tested.
However, even absent benefit from creating new models, there are still good reasons to support the economic changes that would allow such models to be created in Europe:
We need a more unified, EU-wide, approach to investment both private and public; we need to upgrade the power grids even just because they're old and a poor fit for existing supplies and demands let alone forecast new demands and new supplies; and we need local chip factories to protect against supply chain problems in the event of conflict involving existing suppliers.
I expect the AI-training investment bubble to burst before the EU gets any of those things, those things are good regardless.