Citation-infusion of research papers with AI?

I came across the AI company Sourcely these days. It promises to “Finish Your Research in Minutes. Save Your Sleep. Paste your essay to find, summarize, and add credible sources. (That’s something Google Scholar can’t do!)” Saving time for sleep is good and if it would work, it would outperform GS, or any traditional search for citations and references, for that matter. However, I have some thoughts and concerns.

It may be obvious: Finishing research in minutes by infusing citations is an illusory promise. If one had done the research project properly, one would not need a tool (AI-based or not) to gather some sources and insert them in the manuscript at the very end. One should know relevant sources and references at the beginning, or try to find them early on in the project before starting to write. At least, this is the sequence I would recommend.

I pasted a title for a hypothetical article, not an entire essay into the text box on the landing page (there is a 300 characters limit in any case). The results were okay, but not better than what I would have gotten from Google Scholar. It may work better with more text, or with more training of the AI, but it does not seem to be there yet.

If it would work as promised, it would bring the cite-something-for-the-sake-of-citing attitude to its peak. Sourcely promises “credible sources”, leaving me unclear what “credible” is supposed to mean. The hits that I got were published articles from non-predatory journals, as far as I could tell. The publications were credible in that sense. However, some seem too specialized for the title that I entered, some were from fields too remote from the field I used in the title (“political science”).

The bottom line is: I am skeptical about the value of such services, even if they worked as intended. Reading is part of the research process and not the final stage. If one looks for citations only in the last step, one has done something wrong early on in the process. This is not to say that all AI should be banned from research or is useless. ChatGPT and other AI assisting in coding are of great value, for example. However, there may be some parts of the research process for which AI is not made; not yet, at least.