Brainstorming hypothetical scenarios usually results in endless discussions about imaginary problems that never come to pass, and completely missing the real problems. This tends to happen regardless of who participates.

agreed. ...I guess why not have people who're better at it or who'd rather spend time on it, do it? Would probably miss most of the real problems, but could catch a few and help steer complex issues. I guess I'm thinking about it as applying the type of mission that OpenAI has[1] on a wider/smaller scale.

1- https://openai.com/about/#mission

The OpenAI people mean well but their their concerns over safety seem a bit silly considering they haven't actually demonstrated any progress toward building a true AGI. At this stage it's the equivalent of worrying about the problems that would be caused by an alien invasion: an interesting intellectual exercise but useless in practical terms.