The output generated by the model can be selectively responded to.
Example:
Output contains 10 sections, user has to clarify section 3 before proceeding with the task. In this case if user were to prompt the general chat to answer the question it would generate context based around it and the next outputs would also include the “cross contaminated” information generated for the question. Instead if we allow the user to target the question in a separate isolated section within the chat that focuses on section 3, it would increase productivity and allow the model to generate more informed and context based responses.