@noracodes there are two related things here:

(1) is that on-call librarians are generally overkill for this sort of job. LLMs are, by design, only effective at broadly-scoped inquiries into generally-available information. librarians can make reference materials for these sorts of queries which patrons can self-service with; you don't actually need an on-call human being to produce the same exact libguide ad-hoc every time. the problem is that, in the current state of the internet, these reference materials are not discoverable.

(2) is that many people do not want to use LLMs to learn, they want to use LLMs to avoid learning (e·g bash script generation so they don’t have to learn bash). librarians would not help in this case since being an expert at finding information doesn’t mean you can or should be doing other people’s homework for them.

i think (1) is solveable and (2) should not be solved. however, capitalism currently finds profit in (2) and not (1).