March 8, 2023
When I was in college, I had a colleague who was several years older, widely read, and was ready to provide an authoritative answer to any question at the drop of a hat. His answers were literate, often inspired, sometimes witty, and inevitably and at times wildly inaccurate. My friend was not widely published, so it was hard to pin him down as having said something wrong, nor was he ever very clear on where his ideas or even his quotes came from. But he was always up to the task of providing an answer when asked.
Now having such a friend posed a challenge. It was easy to ask him questions, but was it worth doing the work to discern if his answers were correct? Should I just cite him and attribute what he said to his as-yet-unpublished manuscript? Or would it have been better to encourage him to put it in his blog so that I could cite it explicitly by putting down the date it was viewed?
Of course, citations play several roles in writing. They provide attribution of origin for thoughts, insulating one from accusations of plagiarism. They also provide support for arguments by an appeal to authority. That someone said it provides evidence that it must be true, or at least that it is believed to be accurate by someone.
On the other hand, if one develops a reputation for being someone who cites unreliable sources, one loses any cogency for one’s arguments and is instead reduced to being just another unreliable source. For this reason, although my friend was entertaining to converse with, and occasionally provided inspiration for what I wrote, he was never someone who merited citation.
By now you will have seen the obvious parallels between my unreliable friend and ChatGPT. While both are enjoyable to put questions to, and while both can inspire further action, neither is a source to stake your reputation on.
When I saw in the Guardian the other day that the International Baccalaureate (IB) had decided to allow students to cite ChatGPT, I decided to investigate the wisdom of this decision. Rather than put the question to my friend of whether the IB should allow students to cite ChatGPT as a source, I decided to put the question to ChatGPT itself. Curiously, ChatGPT seemed to take a more modest, reasonable approach than I would have expected—one that seems to have more academic integrity than the IB’s approach. I share the highlights of my conversation.
When asked if it is a reliable source, ChatGPT responded with a good deal of modesty:
As an AI language model, ChatGPT is a reliable source for generating ideas and providing general information on a wide range of topics. However, it is important to note that ChatGPT is not a substitute for primary research and should not be the sole source of information for academic writing…it is important to critically evaluate and verify the information provided. Additionally, academic writing typically requires a higher level of specificity, detail, and evidence than what ChatGPT may provide. Therefore it is recommended to use ChatGPT as a tool to assist in research and idea generation, but to supplement it with primary research and other reputable sources.
I then asked the more specific question of whether the IB should allow students to cite ChatGPT as a source. ChatGPT basically says no, providing more detailed reasons, but allows that the IB itself should determine the ultimate rules.
While ChatGPT is a powerful tool that can assist students in their research, it may not be appropriate to cite ChatGPT as a source in academic papers. Instead, students should seek out credible sources of information, such as scholarly articles, textbooks, and reputable websites, and cite those sources in their papers….Furthermore, it is important to note that citing a source like ChatGPT may not meet the specific requirements set by the International Baccalaureate (IB) or other academic institutions. Students should always consult their instructors or academic guidelines to determine what types of sources are acceptable for use in their papers.
To clarify, I asked ChatGPT a question about its own reliability, specifically “if I asked ChatGPT if it was a reliable source, what answer would I get?” The response was enlightening. When queried about its opinion on its own reliability, ChatGPT shows a proper amount of skepticism.
As an Al language model, I am not a source of information in the traditional sense, but rather a tool that can assist in generating ideas and providing information on a wide range of topics. My responses are generated based on patterns and relationships in the text I have been trained on, which is sourced from a wide variety of texts on the internet.
While I strive to provide accurate and helpful responses, it is important to note that my responses are not necessarily vetted for accuracy or reliability in the same way that traditional academic sources are. It is always recommended to verify information from multiple sources before relying on it.
Finally, putting things back into the Academic context, when ChatGPT was asked if citing an unreliable source should result in a failing grade, ChatGPT again takes a modest approach, leaving the question to the instructor, but indicating that in general students should carefully evaluate sources for credibility before citing them.
It is generally expected that students will use credible and reliable sources in their academic work, and citing unreliable sources can compromise the quality and validity of their work. Therefore, it is not uncommon for instructors to deduct points or penalize students for citing unreliable sources. Whether a student should receive a failing grade for citing an unreliable source would depend on the specific circumstances of the assignment and the severity of the citation error. In some cases, citing an unreliable source may be a minor issue that can be corrected with a deduction of a few points or a warning from the instructor. In other cases, citing an unreliable source may be a major issue that calls into question the validity of the entire assignment, in which case a failing grade may be warranted. In general, it is always important for students to carefully evaluate their sources and ensure that they are credible and reliable before citing them in their academic work.
In light of the above, should the IB allow students to cite ChatGPT? It comes down to the question of whether ChatGPT is a reliable source. Either it is or it is not. There is no gray area here.
- If ChatGPT is not a reliable source, then clearly it should not be cited by students in their academic writing without additional supporting citations. And if one has additional, reliable supporting citations, there is little reason to include an unreliable source as well. This suggests that ChatGPT should not be cited.
- If ChatGPT is a reliable source, then we need to take its recommendations seriously. The fact that it asserts that “my responses are not necessarily vetted for accuracy or reliability in the same way that traditional academic sources are” suggests that it sees itself as unreliable, and if it is unreliable, then as we have noted above, it should not be cited.
In either case, the only conclusion to be drawn is that ChapGPT should not be cited.
In light of this discussion, the decision of the IB to allow students to cite ChatGPT is surprising, to say the least. What the ramifications of this decision might be will appear in the companion piece by OESIS President and my fellow OESIS Board Member, Sanje Ratnavale.
You may also be interested in reading more articles written by Ray Ravaglia for Intrepid Ed News.
