I was thinking of a similar point, which is some programmers’ (myself included) tendency to obsess over making small tweaks to update their editor/IDE/shell workflow without really paying attention to whether optimizing that workflow to the nth degree actually saves enough time to make the optimization worthwhile. Similarly, a hypothetical AI might be very useful once you understand how to make the perfect prompt, but the time and effort necessary to figure out how to craft the prompt just right isn’t worth it.
I suspect ChatGPT isn’t quite that narrow, however, and I’ve already seen positive returns to basic experimentation with varying prompts and regenerating answers.
I was thinking of a similar point, which is some programmers’ (myself included) tendency to obsess over making small tweaks to update their editor/IDE/shell workflow without really paying attention to whether optimizing that workflow to the nth degree actually saves enough time to make the optimization worthwhile. Similarly, a hypothetical AI might be very useful once you understand how to make the perfect prompt, but the time and effort necessary to figure out how to craft the prompt just right isn’t worth it.
I suspect ChatGPT isn’t quite that narrow, however, and I’ve already seen positive returns to basic experimentation with varying prompts and regenerating answers.