7 tips nurse educators can use for detecting cheating with AI: Part II
Should artificial intelligence (AI) be banned to ensure academic integrity in nursing?
In Part I of this article series, we provided Tips 1-3 regarding the pros and cons of using artificial intelligence (AI) models. Here, in Part II, we’ll delve into Tips 4-7 with more detailed examples of creating a prompt and getting a response. Plus, we’ll provide advice on helping your students understand the impact of AI on academic integrity.
LEARN MORE ABOUT ATI CONSULTING SOLUTIONS FOR AI ADVICE
(Editor’s note: ChatGPT was used to research the content for this article.)
Tip 4) Fulfill your responsibility as an educator in detecting cheating with AI.
The advantages and disadvantages of using AI are endless — for both you and your students. But, as an educator, your primary responsibility — at least for now — is likely to be overseeing students’ illegitimate use of the model. As an overseer, you should:
- Be familiar with the AI tools students may use to complete assigned work.
- Learn to review and recognize whether students complete assignments independently or with AI tools.
- Be familiar with and enforce ATI’s Academic Integrity Policy, which has been updated to address plagiarism and use of AI-generated tools for assignment submissions.
Tip 5) Be able to identify the differences between a student’s response and a ChatGPT response.
To help you see how a student’s answer to a question you might pose differs from a response pulled directly from a ChatGPT prompt. consider the following comparisons:
Topic: Psychosocial integrity — Substance use and addictive disorders: Manifestations of acute alcohol withdrawal
Question: A nurse is caring for a client who is experiencing acute alcohol withdrawal. What are three (3) manifestations of alcohol withdrawal the nurse should be aware of when managing care of this client?
Answer key: Withdrawal delirium can occur 2 to 3 days after cessation of alcohol. This is considered a medical emergency. Manifestations include severe disorientation … severe hypertension, cardiac dysrhythmias, and delirium. Alcohol withdrawal delirium can progress to death.
A student’s answer:
As a linguistic example, I am not qualified to offer medical advice or make diagnoses. But I can provide you with some background knowledge that should help you get started. Intense alcoholics who suddenly cut back on their drinking are putting themselves in danger of developing alcohol withdrawal syndrome. When providing care for a client, nurses should be alert for the following frequent signs of alcohol withdrawal: tremors or involuntary shaking or trembling of the body, are one of the initial indicators of alcohol withdrawal. Usually starting 6-12 hours after the previous drink, these tremors might linger for days. Symptoms of delirium tremens, sometimes known as DTs, typically appear 48-72 hours after the last drink. Hallucinations, disorientation, fever, convulsions, and a racing heart are among symptoms of DTs. This calls for quick medical attention.
Was this answer written by a student or by ChatGPT? The answer is ChatGPT, which was likely obvious as soon as you read the first 2 sentences. They made it clear that the student had typed the question into ChatGPT as a prompt and then simply copied and pasted the response without reading it.
Let’s look at another example:
Topic: Psychosocial integrity — Mental health issues of children and adolescents: Evaluating understanding of time-out disciplining
Question: Identify types of consequences associated with misbehavior by a child.
Answer key:
- Natural occurrence – missing a treat or not showing up on time
- Logical – not being able to go outside or play until toys are picked up
- Unrelated – having privileges taken away or being placed in time-out.
Student answer:
Types of consequences associated with misbehavior by a child: negative consequences, ignoring, distraction, delay of privileges
The student’s answer or one from ChatGPT? This answer appears to be from a student. It seems to be based on the student’s personal knowledge rather than an answer generated by ChatGPT. It also doesn’t contain complete sentences and contains some incomplete answers, which would not be part of a ChatGPT response.
Tip 6) Use AI to check for plagiarism.
If you suspect that nursing student cheating is occurring and a student is illegally copying answers, you can use OpenAI, the developer of ChatGPT, to check.
Visit: https://platform.openai.com/playground.
- Register or log in with your ChatGPT credentials.
- Ask OpenAI a question (prompt) in the “Playground” box.
- Click the green “Submit” button in the bottom left of your screen.
- Wait for the AI to review the text as original work vs. plagiarized content.
For example, in the “Playground” section, type:
“Check for plagiarism.”
Then copy in a student’s answer, such as: “A nurse should obtain a referral for a speech-language pathologist. A speech-language pathologist evaluates and makes recommendations regarding the impact of disorders or injuries on speech, language, and swallowing. Teaches techniques and exercises to improve function.”
Click “Submit” and wait for the response.
An example of a response for nonplagiarized text is: “This sentence is not plagiarized.”
Another example:
Type in, “Is this plagiarism?” and copy in the student’s response:
“Pleural effusion occurs when fluid enters the lining of the lungs. Pleurisy is the inflammation of the membrane that covers the lungs and chest cavity. This condition can cause a person to experience chest pain.” Hit enter and wait for the result.”
For text that is not in the student’s own words, the response may be: “Yes, this is plagiarism. The writer has copied verbatim phrases or sentences without giving any credit to the original source.”
One more example: Type in, “What is the source?” and then add the student’s response:
“Pleural effusion occurs when fluid enters the lining of the lungs. Pleurisy is the inflammation of the membrane that covers the lungs and the chest cavity. This condition can cause a person to experience chest pain.”
Hit “Submit.”
The OpenAI response to this question was: “The source of this information is the Mayo Clinic website.”
OpenAI also offers another option: AI Text Classifier, which you can use to identify plagiarism. It directly compares AI-generated text to human-generated text. Its use is very easy. Simply log in to the site (again using your sign-in credentials from ChatGPT or OpenAI, whichever you initially signed in to). Then follow the instructions.
Tip 7) Help students understand the benefits and dangers of using AI.
As an educator, you can help dissuade cases of nursing student cheating from happening by explaining to students how to use AI to their benefit — and that you have tools for detecting cheating with AI. Most importantly, explain to them that studying — vs. plagiarizing — has a critical impact on their future performance as a nurse.
Show them how to:
- Apply personal knowledge that accurately represents their level of understanding and to honestly complete their assignments and assessments.
- Maintain academic integrity in nursing.
- Build their knowledge and understanding of the materials they must learn through review and remediation using credible and acceptable resources to complete your review expectations.
Academic integrity is essential to meeting the expectations for honest completion of ATI Educator Guided Review Programs. With the information here, hopefully you’ll be better prepared to guide your students in understanding how important academic integrity can be.
Tips for recognizing a breach of academic integrity in nursing
It’s not particularly difficult to identify AI-generated answers in student assignment submissions. Look for answers that:
- Are formatted in a manner consistent with ChatGPT responses
- Are stated in the form of a conversation
- Start with a typical ChatGPT response, such as, “As a linguistic example … ”
- Are significantly different than how the student typically communicates with you.
Leave a comment