Research group Open AI released ChatGPT at the end of November and has since created a buzz among higher education professionals. We’ve dived into a sea of commentary and curated a selection of articles that may inform your approach to instruction and your conversations with colleagues about this new tool.
The articles discussed below represent the opinions of their respective authors and do not represent an official statement or opinion by the college or Chemeketa’s Center for Academic Innovation.
This is an ever-evolving topic and we anticipate offering periodic updates.
Have limited time?
We recommend Ryan Watkins’ Update Your Course Syllabus for ChatGPT or Torrey Trust’s openly licensed presentation on ChatGPT for immediate steps to take and considerations moving forward.
What is ChatGPT?
ChatGPT is a chatbot built on an artificial intelligence system called Generative Pre-trained Transformer (now in its third generation). The underlying programming is similar to that used by grammar and spell checkers, predictive text tools, and language translators. It has been “trained” on an enormous body of text-based information and further trained by human “supervisors” to fine-tune its ability to respond to a given prompt with fluent, grammatically correct content. It is capable of producing multi-paragraph essays, software program codes, and poetry.
What are its Strengths & Weaknesses?
According to Arvind Narayana and Sayash Kapoor (content warning: language), researchers in the department of computer science and the Center for Information Technology Policy at Princeton University, ChatGPT is good at:
- Tasks where it’s easy for the user to check if the bot’s answer is correct, such as debugging help.
- Tasks where truth is irrelevant, such as writing fiction.
- Tasks for which there does in fact exist a subset of the training data that acts as a source of truth, such as language translation.
They urge newcomers to the technology to recognize that “ChatGPT is often extremely good at answering questions. But the danger is that you can’t tell when it’s wrong unless you already know the answer.” While ChatGPT is exceptional at mimicking formulaic writing, it is not capable of distinguishing truth from fiction within its body of knowledge.
In this NPR interview with Professor Ethan Mollick, Associate Professor at the Wharton School of the University of Pennsylvania, Mollick describes how ChatGPT is capable of summarizing large amounts of text such as an academic paper. It can also proofread and improve writing that you provide. However, he, too, highlights a lack of accuracy in the results. He suggests that the user should think of ChatGPT as “a[n] omniscient, eager-to-please intern who sometimes lies to you.”
Dr. Torrey Trust, from the University of Massachusetts-Amherst, has written a series of slides discussing ChatGPT. She emphasizes that the data ChatGPT was trained on dates no later than 2021. When it does not have the information to include in a response, it will make it up, rather than return an error message.
What are the Implications for Writing-Oriented Courses?
Because ChatGPT is capable of mimicking a wide range of writing styles, commentators believe it will have a significant impact on writing studies, with many raising an alarm about how it can be used to cheat. In his essay, AI Will Augment The Writing Process, Not Replace It, writer and instructor Marc Watkins, from the University of Mississippi, shares his belief that tools such as ChatGPT will become standard technology for writers and writing students. He cautions writing instructors to “be proactive in our response and not approach our teaching out of panic and mistrust of our students. What message would we send our students by using AI-powered detectors to curb their suspected use of an AI writing assistant, when future employers will likely want them to have a range of AI-related skills and competencies?”
Beth McMurtrie, writing in the Chronicle of Higher Education, expands upon Watkins’ perspective in her article, AI and the Future of Undergraduate Writing. The professionals she interviewed are also concerned that focusing narrowly on preventing cheating obscures underlying problems with a “transactional” approach to college. Instead, experts express hope that faculty will have the ways and means to demonstrate the value and purpose of writing and craft meaningful assignments that students want to do. They, too, recommend that AI-tools be integrated into the curriculum.
Ann Mills, an English instructor at the College of Marin, has begun to develop resources that “serve as catalysts for inquiry, discussion and collaborative research as we respond to this major change in the kind of writing assistance available to our students.”
What are the Implications for Assessment?
ChatGPT poses a challenge to certain kinds of assessment and particularly online assessment. For multiple-choice exams and short-answer quizzes,ChatGPT has been shown to be able to answer such questions correctly. Ann Mills offers options for instructors in her resource, How Do We Prevent Learning Loss due to AI Text Generators? Likewise, Cynthia Alby, co-author of Learning That Matters: A Field Guide to Course Design for Transformative Education has offered her short-term suggestions for adapting to teaching and learning with AI.
According to an EdSurge article by Daniel Mollenkamp, plagiarism checkers will soon have the ability to detect AI-generated text based on the current technology, however openly available detectors are not reliable and relatively easy to evade with simple changes to the text.
Ryan Watkins, Professor of Educational Technology Leadership and Human-Technology Collaboration at George Washington University, provides guidance to Update Your Course Syllabus for chatGPT, including the need to include discussing ethics with your students.
How Can it be Used as a Productivity Tool?
Ethan Mollick points out that instructors can leverage the technology to ease some of their workload by generating first drafts of syllabi, quizzes, student feedback, and email messages in his article, The Mechanical Professor.
Torrey Trust includes examples illustrating what ChatGPT can do in her set of slides.
Are there Other, Emerging Concerns?
Technologist Autumn Caines asks readers to think through the consequences (content warning: language) of asking students to use ChatGPT in their studies, including concerns around privacy and unpaid labor.
Torrey Trust also cites privacy concerns and unpaid labor as issues to consider before incorporating a ChatGPT assignment into your course.
And David Karpf, associate professor in the School of Media and Public Affairs at the George Washington University, highlights how ChatGPT will not be free after this initial preview period and explores the likely social and economic constraints that will affect its implementation.
If you’d like to try out ChatGPT and other text generation tools, follow these links:
- Pricing: free (for now, described as a “research preview”)
- Pricing: first 4,000 words are free, plans start at $10/month
- Marketed as a way to break through writer’s block with AI as a writing partner
- Pricing: free but paid upgrade version exists
- Focuses on “paraphrasing” and summary generation