There is the question of liability management. If something bad happens due to an edge case in an algorithm, and people die, who is responsible? With an algori

There is the question of liability management. If something bad happens due to an edge case in an algorithm, and people die, who is responsible? With an algorithm, you are placing responsibility on something you cannot blame. Not popular in situations where human lives are at stake. A human operator, with such a responsibility, will develop and sustain skills to handle any edge case.