In my experience, many librarians are skeptical about AI tools. In many cases, I agree with this skepticism, but I do think there some good use cases. This is one simple example. Given a spreadsheet of books, it tries to guess whether each book is in a particular subject area or not (in this case, mathematics, statistics, or data science) based on the title. It is, of course, not perfect, but could be a great tool for a list which is too large to be worth the trouble for human review. In particular, this takes about a minute per 100 books, and this fast speed could be worth decreased accuracy if it means the classification gets done at all rather than not at all. One could also experiment with different models and different setups to get better results.