The November newsfeed heralded the arrival of AI essay writing. AI (or Artificial Intelligence) essay writing recruits online software that sifts through information and generates a thoughtful written analysis. Enter a prompt, and AI can turn out a reasonable essay on everything from utilitarianism to the Krebs cycle…for free.
In truth, AI authorship hardly qualifies as “news.” There is nothing new about it: AI has been generating blog content for online businesses for some time and is largely responsible for the current glut of meaningless search engine results. Enter a query about stain removal, and the top matches feature links to vacuous 1,500-word essays filled with advertisements and zero practical information. That’s AI.
That said, given the right prompts, AI software can truly deliver some lovely, thoughtful college-level writing. Of course, such a claim should require evidence, so we submitted increasingly specific prompts to a free online AI writer and were unfailingly impressed with the results. Consider this response to a query that asked AI to apply Watson’s Carative Factors to a case study:
One way to apply Watson’s carative factors to a case study is to use them as a framework for analyzing a patient’s care experience. For example, suppose we are studying the care of a patient named Maria who has been hospitalized for treatment of a chronic condition. We can use Watson’s carative factors to guide our analysis as follows:
- The first carative factor is “human caring,” which emphasizes the importance of building a caring relationship with the patient. In Maria’s case, we might look at how the nurses and other healthcare providers interacted with her, and whether they were able to establish a sense of trust and rapport. We might also consider whether the care team took the time to listen to Maria’s concerns and address her needs, and whether they provided support and comfort during her hospital stay.
- The second carative factor is “clinical competence,” which focuses on ensuring that the care provided is evidence-based and meets the highest standards of quality. In Maria’s case, we might look at whether the care team had the necessary knowledge and skills to provide appropriate treatment for her condition, and whether they followed established protocols and guidelines. We might also consider whether the care team used technology and other tools to support decision-making and improve patient outcomes.
- The third carative factor is “creative problem-solving,” which emphasizes the importance of being flexible and adaptable in responding to patients’ needs. In Maria’s case, we might look at whether the care team was able to identify and address any challenges or barriers to her care, and whether they were able to come up with creative solutions to help her manage her condition. For example, if Maria had difficulty complying with her treatment plan, the care team might have worked with her to find alternative strategies that would be more feasible for her.
Overall, by using Watson’s carative factors as a framework, we can gain a more comprehensive understanding of Maria’s care experience and identify areas where the care team may have excelled or where there may be opportunities for improvement.
This is concerning. While students have, for decades, had the opportunity to hire a human writer to complete essay assignments, online AI software services can write essays instantly at prices that range from free to about $50 per month. That’s a notable shift from earlier years where contracting with a human writer was prohibitively expensive and time-consuming.
But the game is not over. A few experiments with online AI software services suggest some ways to address AI essay cheating, and interventions will require refining and revisiting course prompts.
1. AI writers seem to have plagiarism issues
When we enter similar prompts into an AI essay generator, we notice extensive use of identical phrases. This suggests that while the first student who uses AI might go undetected, a second student will not be so lucky. Because most plagiarism detection software detects both the unoriginal material and the source, both students may have some explaining to do.
When asked to generate an “Outraged letter about Miriam Abbott,” and then asked to generate an “Outraged letter about Kipp Abbott,” the letters shared many similarities, including this passage:
“This is unacceptable and a gross violation of the healthcare profession’s duty to provide safe, competent and compassionate care to all patients. The fact that (Abbott) was subjected to such mistreatment is deeply . . . “
Although plagiarism checkers have been integrated into many learning management systems, their associated discussion board platforms often go unmonitored. It would therefore be in the best interests of institutions to identify paths for originality checks for all submitted student materials.
2. AI writers cannot navigate proprietary material
Material that is exclusive to your course will be outside the scope of AI expertise. So, for example, asking AI to explain the results of an experiment or observation that was conducted within the context of your course would likely yield un-submittable results. Further, jargon that is specific to a field seems to also raise questions about the software. While it navigated Watson’s Carative Theory well, when asked about the limitation of the Criteria of Adequacy (a term in a Philosophy of Science text), it was less successful, opening with:
“Anatomic pathologists serve as consultants to their clinical colleagues and patients, and pathology reports officially document this communication . . . “
These results suggest that students who leverage AI for essay writing risk submitting nonsensical answers for course assignments.
3. Most AI relies on text searches
Becauses most AI writing services rely on text-based information, integrating course video materials into essay prompts may also deter AI’s ability to pick up and synthesize content for essay writing.
AI continues to develop and improve, and it’s therefore wise to expect these recommendations to change as the technology evolves. For now, though, we asked AI itself to tell us how to address its threats to academic integrity. AI’s suggestion is as follows:
The essay could also explore the ethical implications of using AI to cheat, and argue that students who do so are not only breaking the rules, but also undermining the integrity of their education and the value of their degrees. Additionally, the essay could discuss the potential countermeasures that could be put in place to prevent students from using AI to cheat, such as using advanced AI systems to detect and prevent cheating, or implementing strict penalties for students who are caught using AI to cheat. Overall, the goal of the essay would be to raise awareness about the potential for AI to be used for cheating, and to encourage students to avoid this unethical practice.
Miriam Bowers-Abbott, MA, is an associate professor and Faculty Fellow at Mount Carmel College of Nursing.
