What is the best way get my bot to return responses that are as human like/conversational as possible?

Hi guys, I want to get my bot to respond to users in a way that so closely resembles human text that people would believe they are speaking to another actual human. Do you think this could be done purely with the right prompting using the base openAI api or would I need to finetune my own model? If so, any ideas about what data I could use to accomplish this? Any help is appreciated.

That would break the Terms of Service, the AI should not be used to fool users into thinking they are communicating with a human, they must always be told in no uncertain terms that they are dealing with an AI.