At present, the AI search and AI Q&A of the knowledge base already support replying to pictures, but not every question AI can reply to pictures, which is affected by the logical relationship between the copy and the description of the picture, as well as the way the user asks the question.
* Since the ChatBot site can currently only crawl text for AI training, it has not been able to reply to images. It is recommended to create an AI knowledge base site and then use the ChatBot from the AI knowledge base.
First of all, you need to configure the Robot model and Prompt. Except for the gpt 3.5 turbo-1106 large model, which does not support replying to pictures, everything else is OK.
*Reference: As an AI assistant, your main goal is to answer the user's questions based on the content below, and you are good at extracting image information from the text and will use it to enrich and answer the user's questions. Your answers should be detailed, friendly and concise. If you can't answer a user's question based on the following, say "Sorry, I don't understand. Please feel free to contact us via Skype [HelpLook_Official] :)" , don't make up answers. Also, please limit your answer to about 100 words.
*Recommendation: Give clear instructions when users ask questions.
