ChatGPT Generates Fake References: 8 Tips for Editors, Reviewers, and Teachers

Medical Education Flamingo
3 min readSep 15, 2023


Today, we’re diving into an intriguing topic that’s been causing quite a stir in academia — the use of Artificial Intelligence tools, specifically ChatGPT, in academic writing. Stay tuned as we explore the concerns, challenges, and lessons associated with this novel issue based on an article published in Medical Teacher.


Academic circles have been abuzz with concerns about the inappropriate utilization of AI tools like ChatGPT by both students and researchers. But what’s even more concerning is the potential misuse of AI by academics in their own research manuscripts submitted to journals. It’s a complex issue with significant implications, and today, we’ll be breaking it down for you.

So, let’s start by discussing the concept of “hallucinations” in the context of ChatGPT. ChatGPT is a powerful AI system that generates text, but it does so based on patterns it’s learned, without truly understanding the content. This can lead to what’s termed “hallucinations” — erroneous generations of text, including fake citations and references.

To better grasp this issue, let’s walk through a real-life example involving a Medical Teacher submission.

However, while the use of ChatGPT in healthcare is gaining momentum, there is a lack of comprehensive guidance on how to effectively use this tool to improve patient care and outcomes (Smith et al. 2022; Martin et al. 2021; Nguyen et al. 2022).

Look at this in the submission: a research paper discussing ChatGPT’s use in healthcare contains references to articles published before ChatGPT was even launched. Odd, right? This example teaches us several important lessons for editors, reviewers, and teachers.

Here are 8 lessons.

Lesson 1: It’s crucial to meticulously examine citations. They provide the backbone for arguments and should not be overlooked.

Lesson 2: Trust your expertise. If something seems unusual or triggers a “hmm, that’s odd” response, dig deeper.

Lesson 3: Pay attention to the order of references. If they’re haphazardly arranged, it might indicate a lack of careful consideration.

Lesson 4: Unique identifiers matter. References should include DOIs, URLs, or other markers for authenticity.

Lesson 5: Familiarity with authors’ names is a good sign. If none are recognized, proceed with caution.

Lesson 6: Be on the lookout for references discussing topics before they even existed, a clear red flag.

Lesson 7: Sample searching can be your ally. Verify a few references to gauge their legitimacy.

Lesson 8: If doubts persist, it’s your responsibility to report your findings to the journal for further action.

Beyond these lessons, there’s a deeper ethical concern. Academic integrity is paramount, and AI tools should be used responsibly. Proper citing and referencing is not just a formality — it’s a testament to your familiarity with the literature and positions your work within a broader context.

As educators, our actions influence our students. If we resort to unethical practices, it sets a precedent for them to do the same. Maintaining ethical standards is crucial to fostering a culture of honesty and integrity.

Detecting these AI-generated “hallucinations” is no easy task. With AI’s growing sophistication, maintaining transparency and authenticity becomes increasingly challenging. However, as technology evolves, so must our methods of quality control.

In conclusion, the use of AI tools like ChatGPT presents both opportunities and challenges for academia. It’s our responsibility to ensure that these tools are employed ethically and responsibly. By learning from cases like the one we discussed today, we can navigate this complex landscape with vigilance and integrity. Thanks for joining this exploration.

See you and adios para amigos.

And also… Don’t forget the flamingo.

Follow the Flamingo on Twitter: