Is ChatGPT the End of Writing? Turnitin Believes It Could Actually Boost Students' Writing Instruction

Detection is Just One Part; Tools Also Planned to Help Teachers Save Time and Better Equip Students for the Workforce

Plagiarism-detection software provider Turnitin is putting the finishing touches on its AI writing features for educators; the detection tool is expected to launch for users of the Turnitin platform in the next few weeks. 

Two former teachers now leading the development of Turnitin’s AI writing tools spoke with THE Journal recently about ChatGPT’s implications for K–12 classrooms and why they think it's important for educators to capitalize on ChatGPT — not simply avoid it or ban it altogether.

David Adamson, principal machine learning scientist at Turnitin, and Patti West-Smith, senior director of customer engagement, have been working on Turnitin’s AI writing detection feature and related tools to help educators using their platform to better understand ChatGPT — and show them teachers how to use AI to save themselves time and how to tweak assignments so that ChatGPT cannot earn a good grade on writing homework. 

Adamson, who taught computer science and math at Digital Harbor High in Baltimore, and West-Smith, who worked in public schools for 19 years as a teacher, curriculum supervisor, and principal, both believe that ChatGPT has presented a growth opportunity — or perhaps more like a growth demand — for writing instruction, which they explained at length in the newest episode of THE Journal Insider podcast. Following are excerpts from that conversation (which you can also hear in full on our podcast page or your favorite podcast platform):

No, ChatGPT Doesn’t Mean the Sky is Falling

West-Smith: “I know that over the last couple of months, there's been a lot of Chicken Little style response, you know, where everybody's just the sky is falling. And I think that's a natural reaction to a moment like this. And to the folks that are out there saying is this the end of teaching as we know, or it is this the end of writing as we know it: I would say that teachers know better than that. It's really important to remember that writing is about thinking. So any response to this moment that says ‘let's have students write less,’ I don't think that's going to be the right way for this to turn out. And I think teachers will figure that out very, very quickly. They're good at understanding students. And sometimes it takes us teachers a minute to catch up to them. But normally we do, and once we do, then we know how to move forward and do so in a way that provides students the guidance that they need.”

Adamson: “We need to stay abreast of how the tools are being used and which tools are getting used to collect representative examples of, of that usage. And I think more importantly, adapt to what's appropriate in classroom usage. You mentioned at the beginning, you know, we want to be more than just we're going to show you a number we're going to catch this much off. We're starting with just showing the instructor what's there so they can understand how AI writing is being used in classroom where it might be being used. But we want to add explanation, we want to add next steps and interpretation.”

Meet the Moment, Don’t Ban ChatGPT

West-Smith: “The challenge that we face right now is not only about technology, it is also about teaching. And we have to make sure that the way that we're approaching instruction matches where the world is, and the technology is the indicator of where the world is. What we're really trying to do is partner both the most cutting-edge technology that David and his team are working on with the pedagogy that sort of matches the moment of where we are right now. 

“So our goal is not to say, ‘hey, you use AI, you're bad.’ The goal is to say, ‘here's some insight for you, teacher, so that you understand what's happening in this work, and you can make the appropriate decisions.’ And we think it really is a wrap-around a situation, right? It's not just a technology challenge. It's also an instructional challenge. … If we create assignments that AI can get an ‘A’ on, then we have a problem in the assignment. But teachers need some guidance around that to understand how it is that you actually craft an assignment so that it isn't vulnerable to the misuse of these tools. 

“So we are working to create resources that will support educators in thinking about what they're doing instructionally and how they're responding to these challenges. At the same time, we're giving them the technology tools that will help them to have the insights to really know what is happening (in students’ work). So for example, my team has just released a rubric that is not designed to score student work — instead, it helps teachers reflect on their assessments or assignments — how to craft a prompt or an assignment so that it is less vulnerable to AI misuse, because we know that there are some things that AI is very good at.”

Where ChatGPT Converges with Writing Instruction Pedagogy

West-Smith: “I'm an English teacher by training, so I know that it only takes a couple of writing assignments for you to understand what your student’s voice sounds like. But there are ways that you can create prompts and assignments that really don't leave a lot of room for students to exercise their voice. So we want to push teachers to create assignments that encourage and reward students using their own voice so that AI doesn't sneak into the picture. There is a way to create an assignment or a task so it will actually encourage students to use their own experiences, to use their own voice, to use their own language. There are ways that you can push up the rigor of the critical thinking and reasoning, which AI gets less good at.

“And the truth is the practices (that are being added to Turnitin’s teacher dashboard to help discourage the misuse of AI writing) are not new brand new practices. They're grounded in best practice around teaching writing anyway. The best writing assignments do ask students to use their own voice. The best writing assignments do ask students to use critical thinking and reasoning and are not at the literal level. The best practices around the writing process include things like peer review, and feedback loops with multiple opportunities for formative feedback and growth. 

“Of course I do not want to demonize teachers — there are very good reasons why those best practices are hard to do consistently, reasons it is challenging for teachers to do some of those things that they know to be best practices. This is a place where AI for good can come to play, right? There are ways that AI can cut down on some of the challenges that actually make it hard for educators to adhere to best practice — like how do you handle that insurmountable stack of papers and get it back to students in a timely way? That encourages us to take shortcuts as educators, because that's overwhelming. 

“We know for example, that actionable feedback is one of the most powerful tools to impact student growth, particularly in writing. But it's hard to do, because it takes a lot of effort to write that kind of feedback on each student composition. So teachers need tools that are going to give them back the time that it takes to adhere to those best practices. So I do think this is a moment where we can grow and learn as a field, and get back to some of these pedagogical best practices.”

How School Bans on ChatGPT Will Widen Equity Gaps

West-Smith: “I think [schools banning ChatGPT] opens up real questions around equity, who has access to the tool is a matter of equity. If students can use their personal devices to get around those kinds of bans, then certain students will be privileged in having access to these tools, while other students will not. Certain students have access to these tools instructionally and learn how to harness their power and use them effectively. Those students are being privileged in the world that they're moving into in the future, there are jobs that will literally not be available to the students who have not had the privilege of that equitable access. 

“So when we just slap a flat ‘This is banned’ sticker on it and say, ‘No one can use these tools,’ we have to really ask ourselves, what are we doing when we make a decision like that? Because we are potentially privileging certain people and really holding back other people — and in my heart, I have a real problem with that. 

“I understand the instinct and the fear, and the [knee-jerk reaction] to say ‘We want to shut these down, we don't want anyone using them inappropriately.’ But with time, discussion, and communication, I think most educators will say, ‘We don't want to flat out ban these tools, but we want to put some guardrails around them so that we know how to use them effectively.’ We know how to maintain academic integrity, we know how to encourage original thinking. And we're going to do those things safely — but with an eye toward not privileging certain groups of students and not providing access to everyone.”

Adamson: “Making sure that everyone has access to these tools, making it easy for instructors to understand how the tools are being used to give them guidance on how they can use the tools effectively in their own instruction [is very important]. Could ChatGPT tools be used as an outlining task, or to help students who are struggling with English language to learn how to rephrase their sentences in a more appropriate academic voice? Tthese may be very appropriate uses that are formative and can bring the students who are struggling upwards to be on the same plane as their peers. So I'm excited because we've barely scratched the surface of what we at Turnitin are going to do. ... But I'm always excited about ways that we can bring these tools to the hands of teachers and students.”

A Little About How Turnitin’s Detector Works

Adamson: “So the way our detector works isn't looking at one sentence at a time, it's looking at whole chunks, at several hundred words, at a time. We want to be better about identifying that and saying, you know, maybe there's some distinctive distinguishing characteristic of AI writing that isn't present here. Turnitin as a whole started with originality measurements: Where did your sources come from? How are you using sources in your text. And these days, it's not going to be a literal copy-paste from somewhere on the internet, it's going to be something that's been mediated in some way by technology, whether it's been paraphrased, or has been AI-written.

“Also, we are growing as an AI writing detection feature; this is not our final form. We will make mistakes, and we want to own that upfront: We're not perfect, our detector is going to highlight text that may not be AI written — the instructor is still in the driver's seat, they have to be the one understanding the student in context. Our goal is that our detection features don’t make very many mistakes at all. But if there a category of writing that folks discover that we're missing, that we really need to get better at, I want folks to reach out to us, there will be a link in the detection tool page for educators to say, ‘Hey, this is something that needs improving.’ And that constant improvement is not only important to me as a scientist, but also to the idea of education of being a learner in this space ourselves.”

Learn more about Turnitin’s development of AI writing features and find training and additional resources for educators at the Academic Integrity in the Age of AI Writing on Turnitin.com.

Featured

  • An elementary school teacher and young students interact with floating holographic screens displaying colorful charts and playful data visualizations in a minimalist classroom setting

    New AI Collaborative to Explore Use of Artificial Intelligence to Improve Teaching and Learning

    Education-focused nonprofits Leading Educators and The Learning Accelerator have partnered to launch the School Teams AI Collaborative, a yearlong pilot initiative that will convene school teams, educators, and thought leaders to explore ways that artificial intelligence can enhance instruction.

  • landscape photo with an AI rubber stamp on top

    California AI Watermarking Bill Supported by OpenAI

    OpenAI, creator of ChatGPT, is backing a California bill that would require tech companies to label AI-generated content in the form of a digital "watermark." The proposed legislation, known as the "California Digital Content Provenance Standards" (AB 3211), aims to ensure transparency in digital media by identifying content created through artificial intelligence. This requirement would apply to a broad range of AI-generated material, from harmless memes to deepfakes that could be used to spread misinformation about political candidates.

  • closeup of laptop and smartphone calendars

    2024 Tech Tactics in Education Conference Agenda Announced

    Registration is free for this fully virtual Sept. 25 event, focused on "Building the Future-Ready Institution" in K-12 and higher education.

  • cloud icon connected to a data network with an alert symbol (a triangle with an exclamation mark) overlaying the cloud

    U.S. Department of Commerce Proposes Reporting Requirements for AI, Cloud Providers

    The United States Department of Commerce is proposing a new reporting requirement for AI developers and cloud providers. This proposed rule from the department's Bureau of Industry and Security (BIS) aims to enhance national security by establishing reporting requirements for the development of advanced AI models and computing clusters.