Today I caught myself genuinely arguing with one of the AI tools I built.
I was redesigning a manual process for work, so I ran my roast skill on the plan. The point of the skill is to be contrarian: poke holes, challenge assumptions, and force the idea to survive contact with a sharper brain.
It worked a little too well.
The roast annoyed me so much that I had to write a response to it.
That made me laugh, because it is also kind of the point. These tools are starting to feel less like passive chat boxes and more like thinking environments. You do not just ask them for output. Sometimes they push back hard enough that you have to clarify what you actually believe.
I used to chat with SimSimi for fun.
Now I am getting ticked off at an AI system I designed because it is doing its job too well.
What a time to be alive.