Implications of ChatGPT
Special Edition | CR's DE Newsletter | 6 February 2023
ChatGPT: Is It All Hype or Will It Change Everything?
Yes, ChatGPT is indeed amazing. Its productions read like they could have been written by a human, though they can be a bit boring and formulaic.
Yes, students could use it to do their homework; it answers math word problems, coding questions, physics questions, etc. It writes essays and creative stories and poems. However, it sometimes get things wrong, sometimes seems to "make things up," and sometimes produces clunky citations. If students use it extensively, yes, they will be shortchanging themselves in terms of learning, thinking, and metacognition processes. AI-use-detection tools are beginning to pop up, and so is discussion about whether or not AI-use-detection is the best way to proceed.
Yes, we teachers can use it too. It creates letters of recommendation, syllabi, assignments, and lesson plans; gives feedback on student work; and offers teaching opportunities (critical thinking--in that it gets some things wrong--and the nuances of ethics, for example).
Yes, it will affect equity. It may perpetuate and intensify the digital divide in terms of access. It definitely perpetuates the biases inherent in our current society. It may also help narrow the digital divide because students with less academic exposure can be freed from worry about grammar and correctness; instead they can refocus on content and thinking.
Yes, it will change the way we teach. In the short term, we may want to revise some assignments and assessments. In the long term, its existence calls for us to rethink what's important and valuable about education and about our work as educators.
So we can say this for sure: it's not nothing. Try out ChatGPT for yourself and see what you think (it's currently free, but may not always be free).
Upcoming ChatGPT and Higher Education Webinars
- February 15, 10-11am PT Artificial Intelligence and Higher Education: Where Are We Going and How Are We Getting There? (live webinar, WCET)
Topics of discussion will include: types of generative AI that are most likely to impact higher education, pedagogical issues related to generative AI and coursework, and academic integrity and artificial intelligence. Registration required. CR is a member of WCET; you may have to set up an individual account under the CR umbrella.
- February 24, 11-12 PT Artificial Intelligence Is Here, Now What? (live webinar, WCET)
WCET members are invited to join us for a Closer Conversation to discuss how you are seeing AI show up on your campuses, how AI might change pedagogical practices in your classes, and the academic integrity issues associated with AI. Registration required. CR is a member of WCET; you may have to set up an individual account under the CR umbrella.
---Recommended Resources---
Summary of Current Discussion
- ChatGPT Is Coming for Classrooms. Don't Panic (article, Wired)
"Critical appropriation" of the technology is the best approach for teachers. Banning it or focusing too much energy on "catching" students using it to circumvent learning will be ineffective, may contribute to shifting focus from learning to grading, and may exacerbate the digital divide. ChatGPT's advent spotlights what might already be "broken" in education; the called-for deep reconsiderations of educational values, systems, and processes are exciting but may look exhausting to teachers fresh from the paradigm shift necessitated by the pandemic.
- Will ChatGPT Change the Way You Teach? (blog, Chronicle of Higher Ed)
This post summarizes much of the ChatGPT discussion and includes links to several resources. The answer is "yes;" you have some options, but ChatGPT will probably change something about the way you teach.
- ChatGPT Advice Academics Can Use Now (article, Inside Higher Ed)
Implications for Higher Ed
- AI and Higher Ed (47:03 podcast, WCET Frontiers Pod)
- ChatGPT: A Must-see Before the Semester Begins (article, Faculty Focus)--included in 1/12/23 DE Newsletter
In the short term, an hour spent exploring ChatGPT may lead to some course revisions before the semester begins. In the long term, it may encourage us to spend more time on helping students develop the information literacy, teamwork, research, study, and metacognitive skills they will need to activate intrinsic motivation and engage in the self-directed learning that this technology can unlock.
- AI Tools like ChatGPT May Reshape Teaching Materials -- And Possibly Substitute Teach (article, EdSurge)
AI can help teachers generate materials to engage students (for example, a video made by AI software from Prof Jim with an avatar of Ada Lovelace explaining Python), reducing workload and/or helping bring content to life in new ways. (Want to try it out? Prof Jim will allow you to create one video for free.) Institutions, though, "need to have their needs and priorities clear" before investing in these new tools and before making decisions about how to employ them.
- What Might ChatGPT Mean for Higher Education? (57:54 virtual forum recording, Future Trends Forum, Academia Next)
The free-ranging discussion in this forum captures many of the ChatGPT issues, especially big-picture issues. Don't have an hour to watch? Highlights include:
- ChatGPT is incredibly easy to use and is currently free. It will not always be free, but it may be quite affordable (responses to prompts cost less than $0.04 each to produce).
- ChatGPT can, on request, create rough drafts and outlines of the writing it has produced.
- It can be used to teach students critical thinking, since it does "make things up" and it does get some things wrong.
- ChatGPT can provide substantive feedback on written work; students can use it as a writing tutor.
- ChatGPT perpetuates the biases that currently exist in our society.
- We may need to start teaching students how to use ChatGPT effectively, since it's not going away as a tool. This skill is likely to be useful in the work world.
- ChatGPT is emphasizing and accelerating the need for higher education to modernize. Now, professors need to focus on developing mentoring and professional relationships with students and introducing them to and helping them navigate academic communities and resources.
- We faculty must become AI literate in order to properly motivate students to use the tools effectively and ethically.
- Probably just about everything faculty do day-to-day right now is not going to be that valuable moving forward, and our roles are likely to be strongly questioned. We need to ask ourselves, what can we do better than the most advanced AI? It's not delivering content or curriculum, it's not assessing student learning, it's not preventing cheating. It is how human we can make our relationships with students: how genuine, how deep, how meaningful.
Implications for Academic Integrity
- How to Cheat on Your Final Paper: Assigning AI for Student Writing (article, AI & Society)
A writing professor at N.C. State University, asked students to cheat on a final essay using an earlier version of GPT, then describe their experience and decide whether or not using AI to write an essay constitutes plagiarism. Rich discussion ensued and important questions emerged.
- Can Anti-Plagiarism Tools Detect When AI Chatbots Write Student Essays? (article, EdSurge)--included in 1/12/23 DE Newsletter
An actual student shares, anonymously, how he has written several essays in collaboration with ChatGPT. TurnItIn is confident that plagiarism checkers will be able to detect ChatGPT usage as soon as this year. Some educators are using this as an opportunity to rethink and more deeply humanize online learning.
- Is ChatGPT Writing Your Students' Homework? A New Technology Will Be Able to Detect It (article, THE Journal)
TurnItIn has released a sneak preview of its AI-writing-detection capability. It has not yet been released, but is in the works. Article includes a ChatGPT-generated literature essay.
- When AI Is Writing, Who Is the Author? (article, Inside Higher Ed)
Though one has to dig a little to find it, ChatGPT has a Sharing and Publication Policy that requires the tool's recognition as a co-author when its work is published. It also places responsibility for following policies on not sharing work that is hateful or that causes social harm on the primary, human, co-author. ChatGPT is an opportunity to complexify the concept of sources and acknowledgement.
Implications for Diversity, Equity, and Inclusion
- Worried about ChatGPT? Don't Be (article, Inside Higher Ed)
While ChatGPT can write grammatical and proficient prose, its voice is mechanical and the work it produces lacks élan and fails to spark curiosity in readers. Given that its prose is boring, ChatGPT's biggest benefit to students may not be allowing them to avoid doing homework and assignments (and learning), but offering them the opportunity to level the playing field in terms of inequities: it can free students from less extensive academic backgrounds from worry about grammar, structure, and convention. And it may therefore play a role in shifting the focus of our teaching of writing, as "students will only gravitate to chat bots if the message they are getting from their writing instructors is that the most important qualities of writing are technical proficiency and correctness." That rethinking of writing as a discipline may also lead to greater equity.
- In an AI World, Let Disability Access Lead the Way (article, Inside Higher Ed)
Instead of thinking of disability as an afterthought, let's prioritize universal access. This focus on AI is an opportunity for proactive equitable access and inclusion. To do this, we need to include representatives of the disability constituency in policy development and all other AI-related discussion.
Practical Suggestions
- Teaching Actual Student Writing in an AI World (article, Inside Higher Ed)
One teacher's suggested strategies to prevent student use of AI: make a policy, get familiar, take a class field trip, require course-based research, unplug, use the testing center, assign content behind the paywall, ask students to show and tell, dispatch students to the archives, and quiz students on their own work. Details about each suggested strategy are included.
- How Do We Prevent Learning Loss Due to AI Text Generators? (crowd sourced document)
This crowd-sourced list includes suggestions on two topics: *Types of essay prompts that text generators may not currently be good at and *Practices and tools that may (or may not) deter text generator abuse.
- Designing Assignments in a ChatGPT Era (article, Inside Higher Ed)
Several teachers grapple with redesigning their assignments and assessments in light of ChatGPT. Some strategies include requiring the use of AI, asking students to engage in a Socratic dialog with ChatGPT, and building a course that an AI could not pass. In his Sentient Syllabus Project, one college professor has created an interactive mindmap (go ahead and click around) as a guide to building a course that only a human could pass (at least for now).
For Writing Teachers
- With ChatGPT, We're All Editors Now (article, Inside Higher Ed)
In the future, much of writing production is likely to be the purview of AI. That means humans will be needed as critical readers and editors. So this is what we should be teaching our writing students--how to be critical readers, how to check the accuracy of sources, how to refine written work with specific purposes and audiences in mind.
---Extensive (But Not Exhaustive) List of Resources---
Articles
- Chat GPT Is All the Rage. But Teens Have Qualms about AI (article, EdWeek)
A national survey of over 1000 teens (13-17 years old) reveals that two thirds of them are concerned about AI's impact on their ability to find a job in the future. A separate survey found that 68% of students think that ChatGPT can help them become better students, and 80% would rather have a teacher who incorporates such technologies into their teaching and learning than a teacher who is scared of the technology.
- Deconstructing ChatGPT on the Future of Continuing Education (article, Inside Higher Ed)
The use of AI in education opens space for new pedagogies, such as heutagogy, or self-determined learning. And tools such as ChatGPT will become learning assistants.
- Freaking Out about ChatGPT--Part I (article, Inside Higher Ed)
Yes, ChatGPT can produce relatively-high-quality writing in response to just about any prompt. Yes, students could use it to cheat. Here are some suggestions that both discourage cheating and deepen learning experiences for students: *Create assignments that are of deep intrinsic interest to students. *Consider artifacts demonstrating processes and experiences when assessing. *Ask students to create metacognitive reflections on their learning. *Revise rubrics so that fluent but dull prose will not result in a high grade. *Integrate this technology into the learning task.
- Guest Post: AI Will Augment, Not Replace (article, Inside Higher Ed)
A teacher using ChatGPT as part of his lessons says that his students are not using the tool to cheat, but to augment their writing and enhance their writing process. The tool may even help unlock greater creativity. Besides, "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 to them to have a range of AI-related skills and competencies?"
- The College Essay Is Dead (article, The Atlantic)
The author looks forward through a forced reimagining of learning in the humanities to a time when AI's expanding exploration of natural language reinvigorates the flagging relevance of the humanities and bridges the divide between the humanities and STEM.
- How to Easily Trick OpenAI's Genius New ChatGPT (article, Fast Company)
ChatGPT is amazing, but not infallible. The article explores some areas of weakness, including esoteric topics, factual accuracy, inability to counter bias, and currency of information.
- Machines Can Craft Essays. How Should Writing Be Taught Now? (article, Inside Higher Ed)
Is increased monitoring the answer to students who "cheat" on their essays by using AI bots to generate often high-quality writing undetectable by plagiarism software? Or must we embrace a future in which such technology is used ubiquitously in the workplace, and introduce students to its effective and ethical use? Or is this an opportunity to dive deeper with learning, such as when the calculator infiltrated math classes? When asked, one AI writing tool recommends that students not use it when learning, but student opinions and experiences are varied.
- NYC Education Department Blocks ChatGPT on School Devices, Networks (article, Chalkbeat New York)
Citing security concerns, academic integrity concerns, misinformation concerns, and offensive language concerns, the New York Education Department has banned ChatGPT on all school-owned devices and school-related networks.
- How About We Put Learning at the Center? (blog post, Inside Higher Ed)
Sure, we can freak out about ChatGPT and we can engage in the cat-and-mouse game of policing student usage. Or we can recognize that ChatGPT's emergence is just spotlighting underlying issues that have needed addressing for a long time, including the issue that while the fundamental purpose of education is to learn (skills, content, critical thinking, creative thinking, etc.), our current system of education does not necessarily support that goal.
- Teaching Experts Are Worried about ChatGPT, but Not for the Reasons You Think (article, Chronicle of Higher Ed)
ChatGPT is only a threat if higher education has become transactional, if students have lost sight of the value of learning, and if teachers are so overworked that they don't have the time or energy to engage or motivate their students. Students invested in their learning and who understand how misusing the tool will undermine that learning are less likely to abuse it. We need to help them make informed decisions about ChatGPT, and to give them assignments worth doing, and help them take ownership over every step of the learning experience.
- Is AI the New Homework Machine? Understanding AI and Its Impact on Higher Education (blog post, WCET Frontiers)
Post includes relevant foundational definitions and sets the stage for WCET's ongoing exploration of the implications of AI on higher ed.
- Ethical College Admissions: "I Am Not a Robot" (article, Inside Higher Ed)
Using ChatGPT to address admissions essay prompts yields mixed results, but it is not easy to identify which essays were written by people and which were written by bots. The article explores potential impacts for college admissions.
- ChatGPT: Threat or Menace? (article, Inside Higher Ed)
We must leverage, rather than resist, the potential that ChatGPT has to impact the status quo of education. And the only way to do this is to understand its strengths (listing resources, calculating, generating starter text, debugging) and its weaknesses (formulating an opinion, judging sources' trustworthiness, departing from formulaic writing, acting on knowledge).
- How ChatGPT Is Redefining Human Expertise: Or How to Be Smart when AI Is Smarter than You (article, Forbes)
AI is better than humans at some things. This means we need to refine our expertise in the things at which humans are better than AI, namely: 1. Knowing what questions to ask; 2. Knowing more than ChatGPT; 3. Knowing how to turn insights into actions. As we continue to automate our thinking, what are we relegating ourselves to? We can push ourselves to be more human, or we can find ourselves acting more like machines.
- ChatGPT Both Is and Is Not Like a Calculator (article, Inside Higher Ed)
The advent of calculators necessitated a reconsideration of mathematics education. In this sense, AI is similar, though with broader disciplinary application. However, calculators allow students more time for mathematical thinking by offloading mechanical operations, whereas AI may deny students one of their exploratory-thinking-stimulation processes (ie: writing).
- GPT in Higher Education (article, Inside Higher Ed)
Reportedly, as one of the investors in ChatGPT, Microsoft has plans to integrate ChatGPT into Bing, its search engine, as early as March 2023. This may mean that students using Bing will be able to generate text. AI is being utilized by multiple apps; banning it is already impossible. As our students will inevitably be expected to use AI in their careers, we must ensure that they develop fluency with the tools. We must adapt.
- Coping with ChatGPT (article, Inside Higher Ed)
A long-time educator's journey through the tides of academic integrity issues, focused on the newest challenge: ChatGPT. Includes a syllabus statement for the present moment, and an optimistic look towards future impacts on education.
- Thoughts about the Impact of AI Text on Assessment (opinion article, Open EdTech)
Cutting-and-pasting from the internet has been going on for quite a while; ChatGPT isn't much different. Rather than trying to ban or avoid AI, we can fix this ongoing assessment problem by creating rethinking educational structures and processes, including "very small classes and frequent discursive engagement." And by recognizing the need for self-expression, which cannot be met by AI text generation.
- Seeing Past the Dazzle of ChatGPT (article, Inside Higher Ed)
One educator's evolving relationship with ChatGPT through a borderline inappropriate crush, then a realization that LLM (Large Language Model) AI bots are merely mimics and are incapable of verifying or judging accuracy of information before using it, then a worry about how its use might disrupt the beneficial "slow thinking" that is accessed when employing the writing process. She arrives at the determination that all decisions we make about how to deploy ChatGPT in education need to be purpose driven: what ends will it serve?
- A Message to Students about "The Bot" (article, Inside Higher Ed)
Writing is an important way to explore and refine one's thinking. If students depend on "The Bot" extensively, they are denying themselves that opportunity to develop and practice their thinking processes and metacognition. Before long, it will begin to be obvious who used "The Bot" to shortcut their academic degrees and therefore can't think as well or as deeply, and who did not.
- Getting the AI We Deserve (article, Inside Higher Ed)
ChatGPT appears to sometimes "make things up." While it's not actually making things up, it is drawing from a huge bank of information, some of which is disinformation. This perpetuation of disinformation is problematic, and educators have a role to play in helping students become better readers in general, and better readers of ChatGPT outputs specifically.
- Rethinking Research Papers, and Other Responses to ChatGPT (article, Chronicle of Higher Ed)
One teacher's revision of the research paper assignment to help avoid tempting students to misuse ChatGPT. At one school, most of the ChatGPT-related violations of academic integrity so far are about students' use of the tool to code, not to write.
Videos & Podcasts
- Teachers, Try This: Build a Lesson Plan Using ChatGPT (3:48 video, Education Week)
Though geared for K-12 teachers, this video explores the potential of ChatGPT to lighten the workload of teachers, allowing them to focus on interacting with their students.
- Expert Warns New AI Chatbot Could Lead to "Nightmare Scenario" (2:52 video, CNN)
ChatGPT convincingly presents incorrect information, which can lead to reactive civil unrest. It may also lead to customer service job loss.
While ChatGPT can pass major exams, is useful in law applications, and can sort through major data banks in minutes, it can't yet be trusted at this stage in its development. The evolving AI-detectors cannot spot misinformation. We're about 75 years away from AI reaching its full capacity; in the meantime, we need legislation to regulate its use and protect us from its misuse.
Interactions with ChatGPT & AI-use Detectors
- My First Chat with the Bot (article, Inside Higher Ed)
The author reports the actual conversation he had with ChatGPT, and comments on its responses. The conversation is free-ranging, involving quotations from literature and music, personal queries, large controversial topics, and an invitation for introspection.
Over the winter break, a Princeton computer science major built an app, GPTZero, which can detect whether or not a piece of text was written by a human. The app uses two indicators, "perplexity" (higher complexity is more likely to be human-written) and "burstiness" (more sentence variation is more likely to be human-written). It is not infallible, but has a significantly successful track record.
- Our Obsession with Cheating Is Ruining Our Relationship with Students (blog post, Marc Watkins, Substack)
The author generated some text with ChatGPT, then ran it through several detectors claiming to be able to identify AI-generated text. All of them declared that the text was likely human-written. The cheating detection cat-and-mouse game is unwinnable. But why try? The vast majority of students want to learn, not to cheat, so let's trust our students and develop learning experiences that focus on skills development.
- AI Writing Detection: A Losing Battle Worth Fighting (article, Inside Higher Ed)
AI-detection tools are not perfect, and may be better used not as "gotcha" tools but as an opportunity to explore what makes human writing unique, as a springboard for discussion, or as a learning tool.
Lists of Resources
- AT Text Generators: Sources to Stimulate Discussion among Teachers (resource list, WAC Clearinghouse)
Phew! There are a ton of resources out there! This list has a teaching-writing focus, and includes some formal and informal (Reddit, TikTok) posts either quoting real students or created by real students.
- Resources for Exploring ChatGPT and Higher Education (resource list, Academia Next)
This list of resources is curated by the host of the smart virtual forum shared above. It includes some podcasts and links to more resource lists in addition to several articles.
- gpt-List (resource list)
A regularly-updated list of products built using GPT. Includes study aids, writing assistants, coding assistants, chat and q&a bots, games, email assistants, ad generators, research engines, legal assistants, language assistants, content repurposers, data analysis assistants, to do list generators, tweet generators, gift recommenders, podcast recappers, and AI-text detectors.