Not really customization but routinely I'd asked ChatGPT to provide side by side comparison table of the two/three/etc of items/elements/technology/etc and it works wonders for understanding, output conciseness and brevity. If you are familiar with the topics it'd be much better if you ask ChatGPT to include any necessary, mandatory or relevant metrics/fields/etc.
For proper prompt customization I personally believe that being a stocastic and non-deterministic NLP approach, LLM needs to be coupled with complementary NLP deterministic approach for example Feature Structure [1]. Apparently CUE is using this technique for its operation and should be used as constraint basis to configure and customize any LLM prompt [2].
For proper prompt customization I personally believe that being a stocastic and non-deterministic NLP approach, LLM needs to be coupled with complementary NLP deterministic approach for example Feature Structure [1]. Apparently CUE is using this technique for its operation and should be used as constraint basis to configure and customize any LLM prompt [2].
[1] Feature structure:
https://en.m.wikipedia.org/wiki/Feature_structure
[2] The Logic of CUE:
https://cuelang.org/docs/concept/the-logic-of-cue/