ChatGPT is the acronym for Chat Generative Pretrained Transformer, created by OpenAI and released in November 2022. It is an advanced chatbot that uses artificial intelligence (AI) and natural language processing to respond to questions. It also responds to requests to create text or images using models trained on information from the Internet, books, articles, and other sources (ChatGPT, 2022). Concern in the academic and scientific communities about ChatGPT is valid and growing. Intelligent.com surveyed college students to determine the prevalence of ChatGPT in homework assignments. The survey found that 30 percent of college students had used ChatGPT for homework. Most of those who used ChatGPT reported using it on more than half of their assignments, even though they believed doing so was cheating. Students assumed that only a few faculty knew about ChatGPT and using it was worth the risk (Intelligent.com, 2023). As of January 2023, four publications in scientific literature listed ChatGPT as an author (Stokel-Walker, 2023). Concerned about undetected ChatGPT-generated text in the scientific literature, Gao et al. (2022) asked ChatGPT to generate abstracts based on 10 article titles from five high-impact journals (50 total abstracts). The investigators then asked scientists to differentiate the AI-generated abstracts from the original abstracts. The scientists correctly identified only 68 percent of abstracts as being generated by ChatGPT. Clearly, ChatGPT is a disruptive technology that will change scientific and educational processes (Marche, 2022). Nurse educators are concerned about the erosion of students’ accountability to learn rather than take shortcuts. ChatGPT generates papers, protocols, and answers to calculation questions; finds errors in computer code; and so much more. Even if two people were to request the same information from ChatGPT, the chatbot would create unique responses. Figure 1 shows text generated by ChatGPT to my request, “Write three different paragraphs about why I want to be a nurse.” Because responses are unique, current plagiarism detection software programs will fail. When one technology emerges, others mimic the original, improve on the capabilities, or counter the original technology. A college student, Edward Tian, wrote a new application called GPTZero to distinguish human-written and machine-written text (Rosalsky & Peaslee, 2023). As an experiment, I uploaded my manuscript and the paragraphs in Figure 1. GPTZero correctly identified the writer of each piece: I wrote the manuscript, and ChatGPT wrote the essays. GPTZero could be a helpful tool for educators to deter students’ use of ChatGPT for assignments.Figure 1: Paragraphs generated by ChatGPT.Not only is the possibility of plagiarism and machine writing a concern, but students’ abilities for conducting rigorous literature searches, synthesizing evidence into their own words, and attributing the work of others can be underdeveloped when ChatGPT is overused (O'Connor & ChatGPT, 2023). Scientific writing could be at risk if future generations of students skip the active writing process by using ChatGPT to write their papers, theses, or dissertations (O'Connor & ChatGPT, 2023). ChatGPT could make the writing process obsolete, just as searches of electronic databases replaced card catalogs. On the other hand, some educators see ChatGPT as a tool comparable to search engines, editing software, statistical software, and reference management systems. For example, nurse educators encourage students to edit their papers using grammar tools in word processors or external tools to improve the quality of their writing. Some educators believe that students can use the formulaic writing produced by ChatGPT as a draft paper. Students would need to check the accuracy of the information by finding and citing source materials properly and add arguments supported by evidence, including research results, case studies, and expert opinions. Thus, student time could be spent on learning about concepts or issues surrounding the topic. Faculty can encourage deep learning by teaching students to critically evaluate ChatGPT-generated information, make informed decisions about the accuracy of the information, and reflect on the use of the tool (Sun & Hoelscher, 2023). For example, a nurse educator might assign students to create an evidence-based protocol on sepsis. Students could use ChatGPT to draft the protocol and use another AI research assistant, Elicit, to find articles that answer their research questions (elicit.org, 2023). Elicit summarizes articles in a structured fashion with an abstract, introduction, method section, interventions, and results. Students could spend most of their time deepening their understanding of the complex pathophysiology of sepsis and the rationales for assessments and interventions. Faculty could grade based on content and meaning making by students rather than on grammar, punctuation, and formatting. Nurse educators should stay current on legislative actions to understand the policy and politics surrounding ChatGPT and other AI products. For example, the US National Institute of Standards and Technology (NIST) is working to create standards for reliable, robust, and trustworthy AI systems (National Council of State Legislatures, 2022). States introduced 17 new AI bills in 2022, with several passing legislation to create advisory groups. For example, the Alabama Council on Advanced Technology and Artificial Intelligence was created in 2021 as an advisory group to the governor (Sharp, 2023). Most bills cite concerns about the protection of private information and public safety as reasons for enacting legislation. ChatGPT is still in its early development. The release last November was designed to collect more data from real users to update its models. The future of chat technologies supported by AI and natural language processing tools is unknown. However, if history repeats itself, this technology is here to stay and will become part of our academic and personal lives.
Karen H. Frith (Mon,) studied this question.