What if they just program it to say Trump won. That's the double edged sword with just telling it what to say. It can be co-opted by bad actors. It also kind of goes against the entire concept of using AI/large language models if you just tell it what to say.