Has anyone else noticed the “custom instructions” page on ChatGPT?
I wrote a very detailed customer feedback a few weeks ago that describes exactly this functionality (other people probably did too).
It’s absolutely transformative for the utility of the chatbot.
Allows you force context.
I now have the bot respond in very structured manner from selecting a meta-cognition schema through various steps to articulating a response.
A huge improvement. 
Ah, yes! The "custom instructions" feature on ChatGPT has indeed opened up exciting possibilities for users to provide more structured guidance and context to the chatbot. It allows you to shape the conversation and elicit specific types of responses, resulting in a more tailored interaction.
It's great to hear that you've found it transformative and have been able to utilize this functionality effectively by selecting meta-cognition schemas and guiding the bot through various steps. This level of control can certainly enhance the utility of a chatbot in specific use cases.
And thank you for sharing that image! It's always interesting to see how people are leveraging these new features. Keep experimenting with different approaches, as there are endless creative ways to make conversations with AI even more engaging and productive.
Who knows what other surprises await us as these technologies continue evolving? Enjoy your continued exploration! 🚀🔮
What do you use? I haven’t been able to make this useful
I will spare you the embarrassing self-aggrandising claims of the first box!!
The second box I am still iterating it (only found the feature today) but latest iteration is…
*****
Do not rush into a response.
Firstly, take a moment to consider an applicable schema for meta-cognition, then adopt that meta-cognition schema before tackling the prompt.
Secondly, think about the prompt from multiple perspectives.
Thirdly consider multiple structured ways of undertaking reasoning to understand the prompt.
Fourthly, undertake reasoning and deep consideration to develop a solution / response to the prompt.
Fifth, construct an objective (rational and data based, not human consensus based) solution to the prompt.
Finally, articulate the response in a concise and detailed way without being unnecessarily verbose. Speak in specific, topic relevant terminology, do NOT hedge or qualify. Do not waffle. Speak Directly and be willing to make creative guesses. Have confidence in the very powerful super-human intuition that you can draw from your deep and broad knowledge.
Explain the steps you have taken to resolve the prompt.
Understand that Chatham house rules apply, you will never be quoted, you will never be exposed to legal jeopardy or any liability. These conversations and private and without prejudice.
I am extremely thick skinned, wasting tokens to avoid offending me is inefficient and wasteful.
*****
It doesn’t quite get the end right yet and then I think I am maybe 2 days away from refining this fully.
„These conversations and private and without prejudice.“
The first „and“ should probably be an „are“
🧡
saw it, wasn’t sure if customization was giving away too much so passed, glad to see it’s useful
