What 50 open source projects taught us about security in the AI era

I have a genuine question. Why can't any of these analyses on security ever throw up a gradual adoption of formal methods? Like, wouldn't that be the next thing that people should slowly start getting used to?