EPISODE 1: THE PROBLEM This post is written by human, me, no AI bullshit here. Let's start with the problem: Many people try to apply AI into automation testing, by having it generating test scripts. As I come across to understand their practice, all I see is desperate script kiddies trying to build house of cards, it is weak, fragile, in short, it sucks. Let's see if you fall into any of these categories: - THE TOKEN BURNER: We all know running AI spends token, and the more you let it consume by reading your tests, understand the codebase of your project, read documents, too many MD files, you are burning unnecessary money. - THE OCD: Some of you chase perfection, something you execute by hand normally, doesn't mean it can be scripted 1-1. Forcing the AI to ensure a passed script is a very bad practice, I will explain later - THE INCOMPETENT BOSS: Let's say you are the boss and AI is your assistant, if you let your AI goes loose, edits anything it wants, deletes anything it deem unnecessary, well that's bad Well it's a lomg fcking post so I will slowly update in other posts