AI’s errors may be impossible to eliminate – what that means for its use in health care

Common sense and the precautionary principle suggest that it is too early for AI to prescribe drugs without human oversight