If we can implement an AI that can watch videos and look at pictures and identify cats, why can’t we have an AI that looks at social media pictures to recognize a (young) person holding a gun or knife?
If we can have NLP parse medical journals and other publications to build knowledge graphs, why can’t we have NLP interpret (simple) social media postings to identify statements posted by an individual with a troubled mind?
We should have an XPrize for developing a service/system that allows schools and counties to opt-in to receive alerts/notifications when their students (current, former/expelled) post content that most rational people would consider to be a telltale sign of troubling (lethal) intent. The community (teachers and students) have a reasonable sense of which students might need help (and should be monitored).
This service/system should make it easy for schools to loop-in their local- and county police departments.