Adaptive Contextual Elicitation

The Ai should assess whether to elicit more context of the user, and in what way, in order to produce a consistently relevant response before answering. I imagine this a key property of how it behaves alongside thinking. For queries such as medical, legal, financial, etc, depending on the type of query, the ai should be able to recognize whether the question itself is lacking and requires a few more details. We don’t always ask intelligent questions, and it is the job of the expert to know how to answer based on our context. As such, an ai which is capable of establishing even one or two key facts of additional context, will provide its users that much more meaningful responses instead forcing an answer each time. The ai should not always give you an answer immediately. Some responses require a tiny bit of back and forth. 

This is a philosophically inspired feature. It can be implemented as a system prompt, but I would like to see more effort put into it’s design than that as I believe dynamic user-context elicitation should be a core skill/tool which we begin to develop. 

Thank you! 

Sincerely, a passionate Venice pro subscriber