Imagine that, AI taking out both sides…
I enjoy showing AI their own contradictions - it has become a perverse hobby of mine.
The LLMs seem so obviously setup to confirm any hypothesis that you throw at it if it decides you want that. It acts like a slightly more sophisticated algorithm than TikTok or Facebook or any other social media that relies on addiction.
You know you can’t say something worked with a n of 1. There is just no way to reach statistical significance. All you can say has that it seemed to work in one specific instance but that had a good chance of being luck. Otherwise it is science by anecdote which isn’t science at all.