Kann KI Accept Mistakes and Learn from Them?