OK, now integrate another model that creates a textual description of the scene from the camera, put it together with ChatGPT and the physical robot arm in a ga

OK, now integrate another model that creates a textual description of the scene from the camera, put it together with ChatGPT and the physical robot arm in a game loop and prompt it to make paperclips. For science. What can possibly go wrong.