Can researchers produce trustworthy AI that serves national