It is difficult to think of a major industry that AI will not transform.~ Andrew Ng (technology entrepreneur)
6 strategies to use when allowing ChatGPT for student writing
In this post, you will find a short list of the pros, cons, and six strategies you can use if you allow the use of ChatGPT for student writing. Professors and teachers at every level are worrying to varying degrees about the impact of ChatGPT in their classrooms. ChatGPT (any version) is an example of generative artificial intelligence (AI) using large language models (LLMs) to generate responses to questions posed by a user. It’s also not the only interactive chatbot out there: Others include Google Bard, Bing Chat, PerplexityAI, JasperAI, ColossalChat, and YouChat. There are sure to be more interactive chatbots coming.
Table of Contents
What should you do as a professor if you allow ChatGPT use for student writing?
- Clearly communicate your expectations to students
- Help students learn how to responsibly use the tool
- Provide guidance on how to properly integrate the chatbot results
- Encourage verification and fact-checking
- Use plagiarism detection tools (not failsafe)
- Offer support and feedback
Using ChatGPT for writing is the focus of this post. The chatbot can be a useful tool for brainstorming, gathering information, and improving language usage, but it can also be abused by students seeking to do little, if any, work and still get a good grade in a class. It seems everyone in academia is currently trying to decide whether it is a tool to help students learn or merely a way to scam the educational evaluation system, with no learning at all for the students.
I’ve experimented with ChatGPT3 and ChatGPT4 with mixed results. In one case, I asked it to take one of my more complex pieces of writing and “make the tone more informal” for speaking. The end result was so simplified that any interesting information was completely scraped from the original. I tried this with both ChatGPT3 and ChatGPT4, and like many other users, found the later version to be “dumber” as verified by Stanford scientists in a prepublication release on ChatGPTs behavior changing over time. I’ve asked for help in creating a book title, and the results were fine, but extremely boring. The titles I came up with myself were a lot more interesting, playful, and provocative. I also asked Chat GPT to write an essay demonstrating a particular point of view, and unsurprisingly, ChatGPT made up references to back the claims.
Pros and Cons of ChatGPT
ChatGPT was released on November 30, 2022, and immediately took off, creating a furor that hasn’t died down yet. Because there is already a lot written about the pros and cons of using ChatGPT for any intellectual endeavor, from science to filmmaking to simple English essays, I’m not going to spend much time on this beyond the simple list of pros and cons below.
Pros
- Access: ChatGPT can retrieve vast amounts of information across various subjects that may not be readily available, or are too time-consuming to obtain, through traditional research methods.
- Generating ideas: ChatGPT can help with brainstorming for a paper, offering suggestions, prompting questions, and helping you think through different angles or approaches to a topic.
- Quick information retrieval: If you need to find specific facts, definitions, or explanations, ChatGPT can provide quick responses based on its training data, pulled mainly from the internet. This can save time compared to manually searching through multiple sources.
- Language assistance: ChatGPT can help improve the clarity and structure of your writing by offering grammar and spelling suggestions, sentence rephrasing, and general language guidance.
Cons
- Privacy issues: For many of us, the lack of privacy of data is a real problem. ChatGPT uploads every conversation, every generative prompt, and any personal information you share for any reason, using it to continue training the AI language models. OpenAI, the American company behind ChatGPT, also notes that there is no provision for deleting specific generative prompts. Fortunately, it is possible to delete any chat either manually or using settings to keep ChatGPT from saving them in the first place or disposing of them after a set time period.
- Lack of verified accuracy: ChatGPT generates responses based on patterns it has learned from its training data, but it does not have real-time fact-checking capabilities. You will need to spend time independently verifying the information provided by ChatGPT to ensure accuracy. This might make it more effort than it’s worth.
- Limited context understanding: ChatGPT can misinterpret the context or intent of a question, leading to irrelevant or incorrect responses. You’ve probably heard the story by now of the author seeking guidance who was told he should leave his wife. It’s crucial to carefully review any generated content and ensure it aligns with your objectives.
- Plagiarism: If you directly copy and paste responses from ChatGPT into a paper without proper citation or attribution, it can be considered plagiarism. Generative prompts can generate ideas and help with understanding concepts, but the content may need to be synthesized, paraphrased, and properly attributed to avoid plagiarism.
- Lack of critical analysis: ChatGPT primarily relies on existing information without the ability to critically evaluate sources or engage in deeper analysis. It’s important to complement ChatGPT’s input with your own critical thinking and analysis to develop a well-rounded paper.
What should you do as a professor if you allow ChatGPT for student writing?
If you are considering allowing students to use ChatGPT for research or papers, below are some ideas you can use to manage it.
1. Clearly communicate your expectations to students
Professors need to clearly state their expectations regarding the use of AI language models like ChatGPT in papers. You should specify whether its use is allowed, prohibited, or subject to specific guidelines. This ensures that students are aware of the boundaries and can make informed decisions. For some examples of syllabi addressing the use of Chat GPT, check out this interactive resource created by Lance Eaton “Classroom Policies for AI Generative Tools.” So far, there are 51 entries, and you can add one of your own if you like. Examples include professors teaching rhetoric and composition who explicitly state “all work submitted must be the student’s own work,” while others are adapting their procedures such any writing to be submitted will be written during class hours.
2. Help students learn how to responsibly use the tool
Professors can educate students about the benefits and limitations of using AI language models. You can emphasize the importance of independent research, critical thinking, and proper citation practices. This may include helping students learn to craft generative prompts. Students need to understand that AI-generated content is a tool, not a substitute for their own analysis and original thought.
3. Provide guidance on how to properly integrate the chatbot results
If you permit the use of ChatGPT or similar tools, provide clear guidelines on how to properly integrate AI-generated content into papers. Some professors are beginning to require that if a tool like ChatGPT is used, the actual prompts used to generate content must be included in the references. The need for attribution, paraphrasing, and synthesis of information, rather than direct copying and pasting, may need to be explained many times. You can also offer examples of appropriate and inappropriate use to illustrate your expectations. The hope is as these tools mature, students (and professors!) will know the process for research and writing, be able to spend less time on citation formats, and more time thinking.
4. Encourage verification and fact-checking
You will need to stress the importance of verifying information obtained from ChatGPT. Students must be encouraged to fact-check the generated content by consulting reliable sources or conducting additional research. Helping students understand the difference between reliable and unreliable sources, and facts versus opinions, will be critical to better quality work. Librarians have been at this for decades, defining information literacy skills as the ability to find, evaluate, organize, use, and communicate information in all its various formats. The next step is to use this information to make decisions, solve problems, or acquire new knowledge. This allows students to develop critical thinking skills and helps ensure the accuracy of their work.
5. Use plagiarism detection tools (not failsafe)
To some extent, professors can utilize plagiarism detection software to identify instances of potential plagiarism or excessive reliance on AI-generated content. This can help maintain academic integrity discouraging improper use of AI language models. Unfortunately, when prompts can include “written in the style of Ernest Hemingway” or “explain the actions of Penelope in the Iliad,” it’s likely that no plagiarism software is going to catch it. The more likely test at the present time is when a student who has struggled mightily to write suddenly becomes extremely fluent and even erudite.
6. Offer support and feedback
Be available as much as possible, given time constraints, to support students throughout the writing process. The most efficient way to do this now may be to provide feedback during class time, especially if the new (old) model is crafting written projects during class time. You may be able to verbally comment on drafts, offer guidance on research strategies, and address any questions or concerns students may have about using AI language models without spending time marking papers. Be curious about what questions the students have and what questions you may ask in return to invoke critical thinking skills.
Future of AI in the classroom
It’s impossible to know how all this hoopla around machine learning/artificial intelligence will eventually play out, in the classroom and our wider society. The Federal Trade Commission has begun an investigation into OpenAI (publisher of ChatGPT) even as CEO Sam Altman testified to Congress that the creators of AI need to be regulated. Many people, including Sundar Pichai, Google CEO, have been discussing the dangers of the various AI releases, not just his own company’s Bard. Microsoft, along with other large software and consulting companies, is struggling to create a “Responsible AI” unit to deal with some of the underlying issues of prejudice and risk embedded in machine learning algorithms.
Education, including the world of higher education, seems to have been caught by surprise at the speed chatbots have been rolled out, as well as their quick adoption by students. There are many universities addressing disruptive technologies, such as the printing press, automatic looms, or the telephone, generally looking at historical patterns for changes in occupations. Yet college classrooms look remarkably similar from century to century, though modern classrooms may embrace the use of YouTube videos or 3D printing for learning. The model of teachers imparting knowledge to students remains largely unchanged.
Meanwhile, students have already incorporated the new technology to write essays and reports on every topic imaginable. Many have already learned the best way to use AI chatbots is to create an outline or generate points to support one point of view. Good students will quickly learn how to add material to make the work sound more like a human student generated it. Some suggestions for dealing with this situation, as indicated above, feel like going back in time: using blue books to write essays in class, doing more oral reports and oral exams, even teaching only the “how” and not requiring the final product such as a research paper, unless crafted inside the class. Students studying language may once again be asked to pick up and read a text aloud and translate it, to prove they are not reliant on translation bots.
Numerous books are coming out now to address AI in the classroom, such as 80 Ways to Use ChatGPT in the Classroom: Using AI to Enhance Teaching and Learning, ChatGPT AI in Education: What it is and How to Use it in the Classroom, Using AI in the Classroom, and The AI Classroom: The Ultimate Guide to Artificial Intelligence in Education for the Higher Education Environment all published in 2023. The 300-page Ultimate Guide offers some interesting suggestions on prompts and editing: The authors describe a general framework for the initial inquiry as P.R.E.P, “Prompt the machine, give it a Role or voice, be Explicit in your instructions, and set the Parameters for the answer. After receiving the generated answer, the next framework offered is E.D.I.T., Evaluate the content for language, facts & structure; Determine accuracy and corroborate sources; Identify biases and misinformation; and Transform the content to reflect any of your own adjustments. This kind of guidance can be immediately helpful to teachers in higher education.
Summary
In the classroom, ChatGPT should be used judiciously, and require independent verification of research and critical analysis. Using some of the ideas and suggestions listed above, you may be better able to navigate the use of ChatGPT by your students to promote responsible academic practices among students without tearing out your hair or throwing up your hands in frustration. For students, ChatGPT is best used complementing traditional teaching and research methods. Given the short attention spans we all suffer from with the current style of entertainment and social media, using chatbots as tools and not expecting them to be a substitute for real learning, could be extremely difficult. I hope we all get better at making the distinction.
Still confused? Contact Hillary for more help and information.
Tags: academia, academic, academic writing, change, communication, planning, strategies, writing
Pingback: Seven ways to celebrate “Ask a stupid question day - Transitioning Your Life
Pingback: 7 ways to train your brain for higher education - Transitioning Your Life