That doesn't explain the many former AI researchers (including Yudkowsky) who quit the field because they came to believe it is a potent danger: they didn't choose to become AI researchers because they started with an assumption that AI research is too dangerous to be allowed to continue.