In this case, does federalism give us fifty laboratories of democracy, where dozens of variations of AI regulation can be tested and compared? Or, since it so hard to identify locations, and to support fragmented models, do we get a de facto national regulatory scheme combining the most restrictive laws from each state?