Hallucinated citations are polluting the scientific literature

https://www.nature.com/articles/d41586-026-00969-z

by 30minAdayHN • 4 months ago

13 3 4 months ago

cratermoon• 4 months ago

Everything about this trend suggests that we’ll soon look back and find we’ve smothered ourselves in the digital equivalent of asbestos

quantified• 4 months ago

Sort of exposes how bad the publication system is. Any literature containing hallucinated references or content wasn't actually generated by scientific practice. Because lack of rigor is so commonplace, the pollution will grow.

adrian_b• 4 months ago

Indeed.

I would never put in a list of references anything that I was not able to read myself, to verify that it really provides appropriate information. It is very frequent to find at other authors references that appear relevant for a subject, so one may be tempted to add them to the list of references, besides those that have really been used. Sometimes they are even claimed by others to be the most important references for that subject, but nonetheless it may be difficult or impossible to access those references. They may have been never digitized, physical copies may exist only in an opposite part of the Earth, or nobody knows where copies may be found.

Whatever the reason, such unverified references known only from hearsay should never be added into the list of references of a research paper.

Unfortunately, as you say, plenty of research papers augment their list of references with many that the authors have never actually seen. Even worse, I have encountered a non-negligible number of cases when the authors claimed that something was written in one of their referenced works, but when I read that myself it said something very different, so it was obvious that the authors had not actually read it, but they just reproduced something from hearsay.

When someone has this habit of adding references that have never been checked, then using an AI for "searching" relevant references takes this bad habit to the next level, by providing not only references that have existed, but can no longer be found, but also references that have never existed.

Hallucinated citations are polluting the scientific literature

Related Stories