> Now, as a matter of academic courtesy it's nice to cite the SE post if it's feasible, but it's impossible for me to remember where I learned everything I've learned over the years. Likewise, an LLM shouldn't be forced to cite where it learned facts, which might be technically hard/impossible, but if there is a clear best source then as a matter of politeness it should. And LLMs mostly do this when they can!
Scientific papers have included references for as long as I can remember. It's not about courtesy, but precision and acknowledgement.
So, are we letting LLMs run free and do things that would be frowned upon if they were done by humans?
When LLMs are used to write academic papers, then there is definitely a stronger requirement on the authors deploying them for identifying the source of ideas, as a matter of professional ethics (not law). But this isn't the issue being discussed, and it's not anything new with LLMs. Long before LLMs, academic authors occasionally got ideas from other places StackExchange without citing them properly in papers. (They frequently learn things from StackExchange, but most of the time they don't need to cite them because it's part of common knowledge. Cases where they got novel, cite-necessary insights is much rarer, as it will be for LLMs.)
Scientific papers have included references for as long as I can remember. It's not about courtesy, but precision and acknowledgement.
So, are we letting LLMs run free and do things that would be frowned upon if they were done by humans?