Might as well be asking if the internet will erode books. No, it will not and does not. AI needs training material and people still need to access webpages to accomplish tasks other than just raw “information gathering”. People read for fun, people want to read opinions, people want to stay up to date, and much more. There are many reasons to read. Yes, it’s all information, but it’s doubtful an LLM will take over all news reporting and journalism.

Anti Commercial-AI license

I have to teach my students to always click the link and go to the source, but I doubt many people do that.

That’s kind of the main thing that bothers me about using these chatbots for research: I’m not going to read the same information twice, unless I really need to verify my sources.

If I do need to verify, reading the generated output is not beneficial for that goal either. So, in both cases, the question comes up why I should read the generated answer rather than the original source.

And in theory, an intelligent AI (beyond LLMs) could reliably interpret the sources and modify the text to fit my question, without halucinations, without misapplying instructions for a different context.
But even if my very intelligent coworkers read a source and tell me about it, it’s still generally helpful to read the source myself, because there is additional context, which they won’t report. And neither can an AI report all that additional context, unless it copy-pastes the original source.