A new artificial intelligence (AI) system called ChatGPT has been released to the public, and many have been shocked to see the extent of its abilities. ChatGPT can accomplish many tasks. For example, it can write poems about any topic, give book recommendations, summarize specific chapters of books, and create workout routines.
If you’re asking whether the AI just Google-searches for responses, the answer is no. If you try to find the source of ChatGPT’s responses, you’ll find they are fresh writing.
This brings us to the most impactful thing ChatGPT can do when it comes to my fellow academics and me. ChatGPT can write reading responses, online discussion-forum posts, short answers, and even essays.
Admittedly, the essays aren’t perfect, but they’re certainly not worse than the worst essays students turn in. Now maybe you’re seeing the issue for education. For years, papers, essays, and other open-response assignments have been the gold standard in preventing cheating. Plagiarism can be an issue, but modern plagiarism detectors and a bit of work on the professors’ side basically eliminate the issue.
AI essays, however, won’t be detectable by these means. Remember, ChatGPT creates original writing. Students will soon have the ability to fake an original essay. And while many academics are horrified by the prospect, I see it as a good opportunity for change in higher ed.
A Necessary Shock
Higher education isn’t the fastest-moving industry. In contrast, consider the restaurant industry. Restaurants have notoriously thin profit margins and open and close at breakneck speed. In your town within the last year, you can probably think of a new restaurant opening, an old restaurant closing, or both.
While you may think this shows the industry is unstable, it’s actually a sign that things are very healthy in the industry. Why? High turnover is evidence of robust competition. Restaurants that fail to satisfy customers close, and high-value competitors thrive.
Put differently, the industry is dynamic. Now contrast that with academe. What is the last high-profile, traditional-university closing you can think of? Certainly, there have been some, but they don’t come quickly to mind.
There are likely several reasons for this, but the most obvious reason is the hand of government. Most public universities can’t go bankrupt because they’re supported by tax dollars. So even if tuition revenue doesn’t cover costs, the public purse does.
But even private universities benefit from government subsidies in the form of student loans. By offering the customers of the traditional higher-education industry loans that would otherwise be unattainable, the government protects the industry from alternative-education competitors.
This lack of dynamism can breed stagnant and counterproductive practices. For example, professors can avoid combative students or simply put less effort into teaching by making courses easy. One way to do this is to make the curriculum easier, but another way is to make the mix of graded work easier.
For instance, a short reading response can be assigned to monitor whether a student is doing the readings. But it doesn’t take long for students to learn that it isn’t extremely hard to “fake” a reading response. By reading a few sentences in the beginning, middle, and end of the reading, or in each section, students can respond to a few ideas without taking on the whole piece.
Professors know this, too, so why not opt for in-class quizzes with specific essays rather than open reading responses? It seems likely that, in many cases, professors would worry about what such a quiz-heavy class would do to their student evaluations.
Online classes utilize another type of assignment, called discussion boards. The boards are notorious among students (even when I was an undergrad) for being mostly fluff, in which everyone agrees with each other to meet the minimum reply quota.
This isn’t to say that those two types of assignment have no place, just that they’re often used as fluff to give the appearance of work to classes.
AI changes this. As AI grows, it will become impossible to ignore the fact that students will be able to score a significant number of these “box-checking” points in one minute by having AI write short essays for them.
These overused easy-point assignments will become increasingly automated, to the point that higher education won’t be able to ignore it. Better practices will need to be adopted.
A Way Forward
So what’s the way forward in this brave new world of AI? Ironically, the way forward is by looking back.
In the history of higher education, homework played a much smaller role. Homework, if assigned, was a way of checking understanding rather than a direct grade determiner.
When homework can be entirely automated by AI, it won’t make any sense for it to be a significant grade driver. It can return to its role of allowing motivated students to check their understanding.
Instead of homework, exams and evaluations were once king, and I believe they will be again.
Oral exams are an endangered species in higher education today, but they are a powerful tool for evaluating knowledge in a world of AI and portable smart devices. Term papers can be replaced, too, by long final exams with open-ended questions that demand handwritten essays.
These sorts of evaluations will be more difficult, and that’s a good thing. High standards drive true learning, and AI will easily be able to destroy the credibility of low-standard evaluation processes.
This also means that professors will likely have to put in more effort in evaluating students. Grading in-class exams and essays is difficult, and increasing the utilization of this method will mean a larger workload. Designing creative ways of assessing student learning that cannot be hoodwinked by AI won’t be easy. But the process of being an effective teacher is not easy.
There are still some negative effects on higher education from AI. Writing classes that involve long research papers obviously cannot make students write a 20-page essay by hand during a three-hour exam window. So certain classes will have new challenges for combating cheating. On the bright side, though, AI still seems a little way from writing high-level research essays.
Overall, though, many types of assignments made obsolete by AI are useless fluff, ossified by an industry that lacks dynamism. I, for one, am happy to see those things go.
This article has been adapted from an article originally published by the James G. Martin Center for Academic Renewal.