The reverse ask is, given unlimited time could an AI tell a school building from a military building, better or worse than a human?
For an example, see the incident described in the article, with context that beforehand the usa had this team of people, and then the usa swapped the people out for the AI, and then this incident happened.