According to Grok, AI actually does get the concept, sorta, but in a very slacker kind of way, and has a tendency to overestimate its ability to manage those consequences. Basically, any consequence that isn't immediate just looks like any other decision tree, and AI assumes it will make the...