When ChatGPT emerged last year, a segment of academia panicked in response to articles suggesting that students could get an AI to write their term paper and you, the instructor, would be none the wiser. Some are ready to write a eulogy for human authorship altogether.

This is a hot topic in academia, and I’m going to try not to retread the arguments about fearing or embracing the new technology. I do want to acknowledge that concerns about academic integrity are legitimate. Aside from the moral hazard of academic dishonesty, there are concrete consequences for students who have not attained appropriate proficiency being passed on to upper division classes or even into the professional world under false pretenses.

But having lived through several similar disruptions over the course of my life, both as a student and an instructional designer, I want to provide some perspective and highlight how education has encountered and integrated new technologies in the past. Each technology is unique and has required different responses from instructors; none are a perfect analogue for generative AI like ChatGPT, but perhaps a reminder of this history will bolster your confidence that we will find a way to ensure our students are actually learning.

• Calculators – Math and related disciplines have perhaps had to deal with technology automating student work the longest, and the capabilities of calculators, both as dedicated devices and as programs within computers and phones, have become more and more capable. In response, it has been standard practice for decades that students are not only evaluated on getting the correct answers, but also demonstrating that they understand the process for arriving at it–to show their work!
• The Internet – I remember in the early days of public adoption of Internet connection in the 1990s, many teachers banned any Internet sources in academic writing, regardless of the credibility of the name behind it. The implication was that if you didn’t physically go to a library, you didn’t actually do research. Even a news article posted word-for-word on a website as it was “in print” was not allowed. Obviously, the Internet has changed since dial-up AOL days. And due to the removal of barriers to publishing information online compared to print, teachers at all levels have had to incorporate information literacy into their lessons to help students differentiate credible from non-credible information from Internet sources. Many instructors and institutions are now leaning into curated and vetted Open Educational Resources from the Internet as a substitute for traditional learning materials.
• Wikipedia – The public, community-edited information repository quickly became a crutch for many students after its debut in 2001. Some instructors quickly moved to ban the use of Wikipedia in academic writing. Eventually, though, most came around to the idea that while a publicly editable platform with no gatekeepers isn’t reliable enough to be cited, it can serve as a helpful resource to become familiar with a new subject and to find more credible sources within page reference lists that could be used in research writing. In addition, we stopped seeing assignments asking students for purely informational summaries of particular topics–in part because those summaries already existed and were easily accessible.
• Google Translate – At its inception in 2006, Google Translate could reliably translate individual words and clumsily translate phrases. Now, for some languages, it can translate longer passages fairly accurately, though not always with nuance or reliability. Language instructors have responded by acknowledging translation tools, coaching students on using them as an aid, and diversifying their assessment practices to include conversational aptitude, rather than just writing.
• Paper Banks – These sites are really an evolution of an older technology: students sharing papers or misrepresenting their authorship on a peer-to-peer basis, rather than in a centralized location. Instructors generally have a sense of work that doesn’t sound like it was written by the student in question, and now resources like Turnitin Originality Checker can show when a student’s work was submitted elsewhere or published on the Internet. It may not be 100% reliable if the original work isn’t in TurnItIn’s database, but it still serves as a healthy deterrent.

In the face of these disruptions, we see a number of strategies, from refining what we value in student work to using other technology to counter potential student cheating. I’m reasonably confident that over time, workable strategies to address AI-created writing will emerge, and will likely be some of the same strategies that have been used to address the above technologies.

Generative AI is an evolving space. It will take time to learn what this technology does well and poorly. It may evolve quickly as ChatGPT improves and other generative AI tools become available. Or we may find hard limits to the kind of information-generation that technologies based on large language models can produce. Today, it may not be realistic to fully ChatGPT-proof your course. That being said, these are the steps I recommend.

Get to know these tools.
What happens when you put one of your assignment prompts, like a quiz question, discussion question, or paper prompt into ChatGPT? Look at the output–what grade would you give it? This can vary widely by subject area and the level you teach. But after a few examples, you’ll likely see patterns. For my own discussion prompts, ChatGPT tended to produce a word salad that contained all the right terms but didn’t connect them in a particularly clear or insightful way. If a student turned this in last year, I might have been inclined to give them a B. Now, I want to rethink my rubric and emphasize clear critical thinking more.

Part of the benefit of this exercise is just acclimating yourself to what AI-generated text looks like. I liken this to movie special effects. The first time you saw Jurassic Park, the computer-generated dinosaurs might have seemed indistinguishable from reality. But with more experience, you can just tell–the cues are there. The lighting and texture aren’t quite right. You might develop a similar sense for AI-created writing that can be a cue to probe when a student’s writing seems off.

Communicate the value of the education, not just the grade.
What is the long-term value of the skills students are gaining as they complete your assignment? If you can answer that question, you can disincentivize students from thinking about academic misconduct. If you can’t answer that question, I’d argue that you have bigger problems than possible cheating.

Focus on Process and Provide Scaffolding
Just as math students are asked to show their work, humanities students and research writers can be asked to provide evidence of their process, from idea generation to research notes to pre-writing to drafting to revision. Tools like ChatGPT can mimic these processes as well, though less convincingly. The main benefit is forcing students to make steady progress on their major assignments to avoid the temptation to take a shortcut at the last minute when the due date is approaching and they haven’t started yet.

Recognize AI’s Virtue as Assistive Technology
Your students may use these tools as writing aids at various stages of their process. They’re good for idea generation, and converting thoughts into specific genres of text. And if the initial output is lacking, students can refine their prompt to get something better. They can copy and paste parts, rearrange things, and make edits. This makes it harder for instructors to identify the influence, but keep in mind that if students can recognize and correct false information in the output, differentiate between good and bad argument, and then edit their way into a better paper, they are learning something!

Plan for Change
For most faculty, it’s not a good use of your time to stay informed on the up-to-the-minute evolution of these technologies, but until the technology matures, plan to check in once per term to see updated recommendations from academic integrity boards and teaching and learning centers based on trends in student usage and evolution of the technology itself.