We are adding OpenAI integration to our educational game engine (Gamefroot). I want to offer our developers (mostly learners) an easy way to add prompts to both static game objects and NPCs in their games.
From an architectural point of view I’m wondering if we should integrate Chat instead of Complete in our node server OR both. I’m asking this question because of this statement in the OpenAI documentation:
Although the chat format is designed to make multi-turn conversations easy, it’s just as useful for single-turn tasks without any conversations (such as those previously served by instruction following models like text-davinci-003 ).
It seems to me that Chat can handle both scenarios, so should be just dump Complete and go with Chat?