Craig Bryan argues that we are overly focused on personalized risk assessments, when much could be gained by focusing on general factors in the environment instead.

Rather than knowing who will attempt suicide and at which moment, we can make all suicide attempts less dangerous.

On the traffic fatality side, we don’t expect that any algorithm will ever be able to tell us when and where any driver will experience a traffic accident, let alone die in a traffic accident, and we generally do not anticipate or expect that any algorithms or tools will ever be developed for this purpose. When it comes to suicide, however, we assume that detection and identification methods are critical or even necessary for prevention, and we generally expect that suicide-risk screening methods and algorithms should be able to do so.