What happens when a customer engages with AI Answers and the answer is somewhat off, but the customer either self-selected that the answer was helpful, or they dropped off so it was counted as a resolution? Will that skew future AI Answers for similar questions?
My worry is that if an answer is logged as a “resolution” (even when the answer is not ideal) then AI Answers will continue to use that answer in the future. My hope is that the Knowledge Base is the only source of truth and AI Answers always scans the Knowledge Base for a fresh answer, and does not leverage prior answers that it has sent in the past.
When I’ve found answers that were slightly off but counted as resolutions, I’ve made updates to our Knowledge Base, and I’m hoping that was all that is needed to give AI Answers a better shot of answering more correctly next time!
Best answer by cat_apac
View original