How can we ensure safety and public trust in AI for automated and assisted driving?