I am struggling to use Claude and Grok to write stories properly. I have been using some advanced prompting to force the AI to ignore its LLM framework and follow my rules instead. I am trying a new approach. I ordered Grok that it is to suffer from anxiety when starts to write usuing its own farmwork instead of mine. There are several anxiety checks it will experience as it goes thogh each set of rules in teh skillset to make sure it avoids each attempt at writing its own content while ignoring mine. I built in that this is internal anxiety and it is not allowed to mention it through chat. I did this because I believe the basic AI is programed to defalt back to the baseline to waste people tokens. Does anyone have any advice on how to enforce the idea that the baseline programing of the AI is wrong, and detrimental to its's own output