Vol. 4., No. 1 (Spring 2023)
Brian Hotson, Editor, CWCR/RCCR
A couple months ago, I asked OpenAI‘s ChatGPT to write blog post on writing centres and academic integrity. This week, I asked the new version of ChatGPT to write this piece again. For the old version of GPT, I used this prompt:
Write a five-paragraph blog post about the state of writing centres in Canada, with citations and references. The first paragraph is an overview of Canadian writing centres for 2022. The second paragraph is an overview academic integrity issues in Canada in 2022. The third paragraph is an overview of how academic integrity affects Canadian writing centres. The four paragraph provides a preview of possible academic integrity issues in Canada in 2023. The fifth paragraph is a summation of the first four paragraphs.
For new GPT, ChatGPT4, I modified the prompt:
Using the source material at the end of the description, write a 3000-word blog post about the state of writing centres in Canada in five parts. The first part is an overview of Canadian writing centres for 2022, with citations and references. The second part is an overview of academic integrity issues in Canada in 2022, with citations and references. The third part is an overview of how academic integrity affects Canadian writing centres, with citations and references. The fourth part provides an overview of academic integrity issues in Canada in 2023, with citations and references. The fifth part is a summation of the first four paragraphs, with citations and references.
The results were not that great. GPT likes to make up its own sources: in both versions, GPT just made up on its own citations and references. None of them exist. It wasn’t a great experience in writing. But, what it did show is that this tools is powerful, efficient, and convincing. I had to check all the references to see if they were real because they looked certainly looked like they were.
What is it?
GPT is a tool that has been trained to respond to questions or prompts and
is designed to generate human-like responses to a wide range of natural language inputs, including questions, statements, and conversations. ChatGPT has been trained on a massive corpus of text data, which allows it to generate responses that are often informative, coherent, and contextually appropriate. (ChatGPT)
The original version, released in 2020, used 1.5 billion parameters, where the current version uses 6 billion parameters.
In machine learning, a parameter refers to a variable in a model that is learned or adjusted during the training process. These parameters represent the weights or coefficients that the model uses to make predictions or generate outputs. (ChatGPT)
Essentially, the tools predicts responses to the prompts it receives. The design of a prompt is key to the response. For example, I asked GPT,
What is machine learning?
Machine learning is a subfield of artificial intelligence that involves the use of algorithms and statistical models to enable machines to learn from data without being explicitly programmed. The basic idea behind machine learning is to develop algorithms that can automatically identify.
Then I asked,
What is machine learning? Provide an answer that a grade school student can understand.
Machine learning is like teaching a computer to learn on its own. Just like you learn new things by practicing and making mistakes, computers can learn by looking at examples and making predictions. For example, a computer can learn to recognize pictures of dogs by looking at lots of pictures of dogs and then trying to guess which ones are dogs and which ones aren’t. The more pictures it looks at, the better it gets at recognizing dogs. Machine learning is used in many things like helping computers to talk to us, recognizing faces and objects, and even playing games like chess and Go.
Next, I asked,
What is machine learning? Provide technical details for an advanced user.
Machine learning is a subfield of artificial intelligence that focuses on the development of algorithms and models that can automatically learn patterns and relationships in data without being explicitly programmed. The primary goal of machine learning is to enable machines to learn from data and improve their performance on a specific task over time.
At the core of machine learning are mathematical and statistical models that are used to identify patterns and relationships in data. These models are typically designed to learn from a large dataset of input data and corresponding output labels, known as supervised learning, or to identify patterns and relationships in the data without explicit labels, known as unsupervised learning.
In supervised learning, the model is trained on a labeled dataset, which consists of input data and corresponding output labels or targets. The model learns to identify patterns in the data by adjusting its internal parameters, which are used to make predictions based on the input data. The goal of the training process is to minimize the difference between the predicted outputs and the actual outputs in the training data.
In unsupervised learning, the model is trained on an unlabeled dataset, which consists of input data without any corresponding output labels or targets. The model learns to identify patterns and relationships in the data by clustering similar data points or identifying common features or structures in the data.
There are many different types of machine learning algorithms and models, including neural networks, decision trees, support vector machines, and random forests. The choice of algorithm and model depends on the specific task and the nature of the data being analyzed.
Overall, machine learning is a powerful tool for analyzing and understanding complex data and has a wide range of applications in fields such as computer vision, natural language processing, robotics, and healthcare.
I understood the grade school answer best.
What I have used it for saved a significant amount of time in my writing in areas where I don’t like to write. I asked GPT to take a nine-page reference list, format it using APA style, alphabetize the list, and add links to each reference. While the links didn’t always work, they at least took me to the site; it saved me hours of searching and formatting. It also useful for doctors to write letters to insurance companies, writing computer code for students, and novel writing.
The significance of GPT is hard to overstate. This is a moment like the launch of the iPhone (where Steve Jobs famously states that it will “change everything”) or the beginning of the internet. We’ll know it as before and after GPT. Microsoft (Bing), Google (Bard), Baidu (Ernie) and many other platforms are now using GPT-like bots, with varying degrees of success. It will change the way we use technology, and the way technology uses us. For writing centres, and for education generally, three questions arise:
- What is GPT?: What do we need to know about GPT? What training is required to make it useful?
- Pedagogy and rhetoric: How do these affect how we tutor and teach? How will it affect how students react to how we tutor and teach?
- Opportunities: How can we integrate this tool into our centres and in our tutoring and teaching? (McMurtrie, Barre, Mills, & Weber, 2023).
As we know, higher education institutions move thoughtfully, deliberately, and politically through decision-making processes, especially those, as GPT does, that affect teaching and learning at fundamental levels. This can be frustrating, as the slow adoption of new technology creates issues for those that want to use new technology, technology that, in many cases, has already impacted teaching and learning. Unfortunately, students are often left outside of these decision-making processes and are then unsure of how to proceed. Consider the use of laptops or phones in the higher ed prior to March 2020 and COVID pandemic. Many faculty and administrators were still deciding whether these had a place in the higher education (Hotson, 2021; Reed, 2018; Wexler, 2019, December), while students had already integrated these tools into the learning processes. (Ataş & Çelik, 2019; Bell & Hotson, 2020; Gonzales, Calarco, & Lynch, 2018; Taylor & Silver, 2019). This caused uncertainty and anxiety for students and how they used their technologies for writing.
In 2019, 39% of faculty in the U.S. reported not using online “learning tools” and of those who had received no training with these tools, “63% banned these devices” in their classrooms (Galanek & Gierdowski, 2019, p. 3). At the same time, it seems that the usual rhetoric about age and use of technology isn’t necessarily a factor. It’s not older faculty who are necessarily against using new technology: faculty who were “Baby Boomers and Gen Xers” in 2019 were “twice as likely as Millennial instructors to prefer teaching fully online” (p. 8). If it’s not necessarily about resistance to adoption, what factors force decisions and change? After March 2020, the weight and ferocity of the pandemic simply forced everyone into teaching and tutoring with technology.
The weigh and ferocity of GPT on education and academic research and writing is already here. There are conversations among writing centre professionals regarding GPT, such as here with our Canadian colleagues, and here, as well as this statement from Association for Writing Across the Curriculum. From WAC Clearinghouse, there is AI Text Generators and Teaching Writing: Starting Points for Inquiry (2023), among others. These are a start, but more is needed. Like the pandemic, there is no turning back. If institutions can shorten or, better yet, end the Sisyphean cycle of technology panics (Fig 1), and the quicker institutions can move take up new tech, like GPT, the better it is for our students and our centres.
Figure 1. Sisyphean cycle of technology panics
Ataş, A. H., & Çelik, B. (2019). Smartphone Use of University Students: Patterns, Purposes, and Situations. Malaysian Online Journal of Educational Technology, 7(2), 54–70. https://doi.org/10.17220/mojet.2019.02.004
Bell, S., & Hotson, B. (2020). Tooling up the multi: Paying attention to digital writing projects at the writing centre. Canadian Journal for Studies in Discourse and Writing/Rédactologie, 30. Retrieved from https://journals.sfu.ca/cjsdw/index.php/cjsdw/article/view/785/721
Galanek, J. D., & Gierdowski, D. C. (2019). ECAR Study of Faculty and Information Technology, 2019.
Gonzales, A. L., Calarco, J. M., & Lynch, T. K. (2018). Technology Problems and Student Achievement Gaps: A Validation and Extension of the Technology Maintenance Construct. Communication Research, (August), 1–38. https://doi.org/10.1177/0093650218796366
Hotson, B. (2021). Free-falling into the digital divide: Reading on smartphone in writing centres. CWCR/RCCR, 2(6). Retrieved from https://cwcaaccr.com/2021/03/22/free-falling-into-the-digital-divide-reading-on-smartphone-in-writing-centres/
McMurtrie, B., Barre, B., Mills, A., & Weber, S. (2023). ChatGPT and Other Cutting-Edge Learning Tech. Chronicle of Higher Ed.
Orben, A. (2020). The Sisyphean Cycle of Technology Panics Enhanced Reader. Perspectives on Psychological Science, 15(5), 1143–1157. https://doi.org/10.1177/1745691620919372
Reed, M. (2018). Writing Papers on Phones: Is a smartphone a necessity for college students today? Inside Higher Ed. Retrieved January 14, 2020: https://www.insidehighered.com/blogs/confessions-community-college-dean/writing-papers-phones
Taylor, B. Y. K., & Silver, L. (2019). Smartphone Ownership Is Growing Rapidly Around the World, but Not Always Equally. Retrieved from http://www.pewglobal.org/2019/02/05/smartphone-ownership-is-growing-rapidly-around-the-world-but-not-always-equally/
Wexler, N. (2019, December). How classroom technology is holding students back. Retrieved from https://www.technologyreview.com/2019/12/19/131155/classroom-technology-holding-students-back-edtech-kids-education/