When the "explain" provides the answer and the answer is not good-enough, we can suggest user to anotate the log instead.
Something like:
"Don't you like this answer, help us train our model by providing correct answer [Click here]" and that will load the log and user will be able to annotate it.