It's far from obvious to me that an AI would be able to inspect a complex system like this for (potentially unanticipated) failures more effectively than a know

It's far from obvious to me that an AI would be able to inspect a complex system like this for (potentially unanticipated) failures more effectively than a knowledgeable, experienced human. And aside from the proximate cause of the fire, the thread highlights a deeper and more insidious issue, namely the diffusion of management responsibility for decisions and their consequences. It seems to me that delegating inspections to an algorithm would make that problem much worse, not better.