It isn't very weird that these smart surveillance thingys go off on the wrong things. What do they have to train those AI? They can't have that many "sound bites" from actual school shootings right? Maybe they actually have some but probably not enough to train a reliable AI on imo. I wonder how they did to create more data. Did they hire some actors and tried to create some "aggressive noises" themselves? I really wanted the journalists to ask the companies this.
