And the errors are really basic, like translating shortly to short, not the same thing at all!

So to me it makes sense to have models with different architecture/data/post training refine each other's answers. I have no idea whether adding the personas would be expected to make a difference though.

But I agree, empirically religion and moral behavior seem at best uncorrelated.

One that I find compelling is that Rawls' veil of ignorance lets us imagine that we might be on either side of a conflict, and that therefore moral actions are equitable to both sides. This gives us a secular morality that doesn't come with the baggage of religious outgroup dynamics.

Also American truckers do a lot more long-haul work and American roads are noisier, so they prefer being more insulated from the engine.

In the UK once website blocking powers were established, their scope was extended repeatedly by courts for IP protection purposes.