This column of news, views and snippets from the international literature of books, libraries, and information, appeared in the March 2024 issue of LIASA-in-Touch, the quarterly newsletter of the Library and Information Association of South Africa.

Since I wrote about AI (artificial intelligence) and ChatGPT in Global Gleanings #16, the literature on it has burgeoned. At the 2023 IFLA World Library and Information Congress, I counted at least eleven sessions dealing with some aspect of AI. I was reminded of the impact of ChatGPT when, in a Zoom meeting with international colleagues, the editor of a respected LIS journal mentioned that he had been receiving article submissions with fake references, i.e., references to sources that do not exist, probably generated by AI.

Chat GPT in academia

Academic libraries play a crucial role in supporting students’ research and writing processes. With the increasing use of ChatGPT by students to write assignments, it is important for academic libraries to respond effectively to this trend. Several studies have explored the impact of AI-powered chatbots on education and student learning. For instance, Smith (2020) found that AI chatbots can enhance students’ writing skills by providing instant feedback and suggestions. Additionally, Johnson (2019) highlighted the potential of AI chatbots in improving students’ critical thinking and problem-solving abilities.

In response to the use of ChatGPT for assignments, academic libraries can adopt several strategies. Firstly, they can provide workshops or training sessions to educate students on the appropriate use of AI chatbots in academic writing. This can include discussions on the limitations of AI-generated content and the importance of critical evaluation. Secondly, libraries can collaborate with faculty members to develop guidelines or best practices for using AI chatbots in assignments. This can help ensure that students are using these tools as supplements to their own research and writing processes, rather than relying solely on AI-generated content.

Furthermore, academic libraries can curate a collection of reliable and authoritative resources on AI and its impact on education. This can include books, articles, and online resources that discuss the benefits and challenges of using AI chatbots in academic writing. By providing access to such resources, libraries can empower students to make informed decisions about incorporating AI tools into their assignments.

In conclusion, academic libraries should respond to the increasing use of ChatGPT by students to write assignments by providing education, guidelines, and curated resources. This approach can help students effectively utilize AI chatbots as supplementary tools while maintaining the integrity of their academic work.

References:

Smith, A. (2020). The impact of AI chatbots on student writing skills. Journal of Educational Technology, 45(2), 123-145.

Johnson, B. (2019). Enhancing critical thinking through AI chatbots in higher education. International Journal of Educational Technology, 32(4), 567-589.

DISCLOSURE

The above section on ‘ChatGPT in academia’ was produced by ChatGPT. I had requested my 18-year old granddaughter (grandkids are more up to date with the latest tech) to submit the following request to ChatGPT for an imaginary student assignment:

Write a paragraph of not more than 250 words on the topic, “How should academic libraries respond to the increasing use of ChatGPT by students to write assignments?” You should include between three and five references to relevant literature, using the Chicago Author/Year citation style.

Read that section again. Not a bad effort, don’t you think?  That is, until you inspect the two references cited in it: Smith, A. (2020). The impact of AI chatbots on student writing skills. Journal of Educational Technology, 45(2), 123-145; and Johnson, B. (2019). Enhancing critical thinking through AI chatbots in higher education. International Journal of Educational Technology, 32(4), 567-589. Alerted by my editor friend, I looked for these articles in vain using Google Scholar. There are various journals devoted to educational technology. Only one bears the title, Journal of educational technology. It is a genuine journal, published in Indonesia, but it has not yet published volume 45. The latest volume published (in 2023) is volume 7. I could find no sign of the article by Smith in it. Similarly there was an International journal of educational technology but it seems to have been discontinued. Its last volume, number 6, was published in 2019. I concluded that Chat GPT had simply made up these quite genuine-looking references. I suppose the two very common family names, Johnson and Smith, were a giveaway.

Arms race or peaceful coexistence?

The use of artificial intelligence in academia is a hot topic in education (Cotton et al. 2023; Teel et al. 2023). It is linked to plagiarism and cheating. Is using AI really plagiarism? After all, no other author’s work is being copied. The output is new and has never been published anywhere. However, it is dishonest to pass off as one’s own work a piece produced by ChatGPT. Cheating is unacceptable. Simply copying text written by ChatGPT short-circuits the learning process that is supposed to take place when students write assignments. In Australia, Tech Business News rather alarmingly warned that ChatGPT “may lead to the downfall of education and critical thinking”, and cited educators’ concerns about “cognitive offloading”, which is “the process of reducing cognitive effort by using external aids.” Like our muscles, our brains need to be exercised to keep in good shape. I prefer not to entrust my financial affairs, my root canal treatment, or a knee replacement to professionals who obtained their qualifications by false pretences.

Not surprisingly, computer boffins have been hard at work trying to outwit this latest threat to academic integrity. Edward Tian, a 22-year-old computer science student at Princeton University, has written an app called GPTZero that can distinguish most of the time (but not always) between text written by humans and that produced by ChatGPT. It uses two indicators: perplexity and ‘burstiness’. Perplexity measures the complexity of the text. Because of the way the chatbot is trained, it tends to produce simpler text. More complex text is more likely to have been written by a human. ‘Burstiness’ refers to the rhythm of the writing. Humans tend to write sentences of varying length: long, complex sentences and short sentences occur. ChatGPT writes more uniform sentences (Bowman 2023). So, professors, if your students are submitting assignments with awkward and complicated sentences, take heart. They are probably doing their own writing. If the text reads like something that has been through the mill of an American mid-range non-fiction editor, with low perplexity and low burstiness, it may have been produced by ChatGPT. And, as we saw above, fake references are a dead giveaway.

Should we worry? An anonymous questionnaire survey undertaken at the Erasmus University in the Netherlands found that 92% of the students used ChatGPT regularly for their studies. However, students were critical of what it produced. They may find it useful for generating and testing ideas, but they do not consider it reliable. If they use text it has produced, they edit and revise it carefully (Smaling 2023). One such student reported, “I always edit it first and then revise it several times. ChatGPT isn’t good enough, and it would be silly because I’m supposed to be learning about subjects myself.” That seems to be a sensible attitude. I hope that was an honest comment!

References

Bowman, Emma. 2023. ‘A College Student Created an App That Can Tell Whether AI Wrote an Essay’. NPR, 9 January 2023, sec. Technology. https://www.npr.org/2023/01/09/1147549845/gptzero-ai-chatgpt-edward-tian-plagiarism.

Cotton, Debby R. E., Peter A. Cotton, and J. Reuben Shipway. 2023. ‘Chatting and Cheating: Ensuring Academic Integrity in the Era of ChatGPT’. Innovations in Education and Teaching International 0 (0): 1–12. https://doi.org/10.1080/14703297.2023.2190148.

Smaling, Elmer. 2023. ‘Almost Everyone Uses ChatGPT, but Students Say They Stop Short of Plagiarism’. Erasmus Magazine. 16 November 2023. https://www.erasmusmagazine.nl/en/2023/11/16/almost-everyone-uses-chatgpt-but-students-stop-short-of-plagiarism/.

Teel, Zoë (Abbie), Ting Wang, and Brady Lund. 2023. ‘ChatGPT Conundrums: Probing Plagiarism and Parroting Problems in Higher Education Practices’. College & Research Libraries News 84 (6): 205–8. https://doi.org/10.5860/crln.84.6.205.