Rethinking the Maturity of Artificial Intelligence in Safety-Critical Settings