The Limitations of ChatGPT: A Cautionary Tale

Imagine the possibilities of having a research sidekick at your fingertips, a tool that could trawl through thousands of journals and produce a list of the hottest articles on a topic. The thought of having access to the latest research with just a simple request is nothing short of thrilling. But alas, as I discovered last night, the reality falls short of the dream.

I asked ChatGPT for current articles on the teaching of primary science and what I received was a list of made-up titles. Even after requesting journal titles, it failed to provide direct URL’s. The disappointment was palpable when I tried to locate these articles and found that they didn’t exist. Google and academic libraries had no record of them.

ChatGPT is an aggregation tool, taking in large amounts of content and remixing it into new content. This is all well and good for creating a story, an essay or a poem, but it falls short when it comes to providing references. It takes a title from here, an author from there, and a date and journal name from somewhere else, mashing them together into a new entity that does not exist outside of ChatGPT.

I can only hope that in the future, ChatGPT will be trained to not do this for references. But for now, it’s an unfortunate limitation that we must face.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *