The Importance of Tutoring Systems in Large Language Models

I am very optimistic about those intelligent tutoring systems you were mentioning in the paper. I know one of the main criticisms people are quick to bring to the table when discussing large language models is their ability to hallucinate. It would certainly be bad for a student to show up to a class after working with a tutor and be filled with all types of incorrect ideas. Do you share this concern? Certainly when we use these systems we have to have the capability to think critically. And that is key here. If we're talking about tutoring systems these are for learning. Students don't have the capability of checking if things are correct because they're learning them. They seem to have worked out something