A framework for better ChatGPT results starts with detailed, specific prompts that constrain the model's output. Then ask it to compare the two best versions of its response. Trust the model when your prompts are precise.
At this point, again, we are following, it seems like it does follow our modes of work framework. So at this point, the next logical move is to constrain the app, right? So I was trying to tell ChatGPT to basically reduce certain things like the highlights, or I was trying to be very detailed about what I wanted. So we follow our modes of work framework. So here I just ask a very Specific question, which is compare the two best versions. So I'm trusting GPT in this case, and I think this is where GPT, along with a functioning app, is really strong.
