Chapter 1: The Reality of AI in the Classroom

It has occurred. One of my students utilized ChatGPT to compose their final paper for my English course. How did I come to this conclusion? Well, I can't say for certain. That's the dilemma.

However, I have a strong suspicion. This particular essay starkly contrasted with the rest of the student’s submissions throughout the semester. The writing lacked personality, coming across as bland and mechanically flawless—definitely not reflective of this student's usual style.

I anticipated this moment would arrive, yet I had hoped to maintain the traditional approach for as long as possible. Even last spring, when I first introduced GPT-3 to my colleagues in the English department, I believed we had a few more years to devise a strategy.

But alas, the moment has arrived.

What action did I take? Initially, I shook my head, then hung it low, and finally scratched it in confusion. I realized that proving this student had engaged in academic dishonesty would be nearly impossible. However, after some online research, I discovered that a machine-learning company named Hugging Face had developed a detector for GPT-2 a few years back. I found it here and tested it with my student's essay. In mere seconds, the detector indicated a 99.98% likelihood that the work was "artificial," rather than "authentic." I subsequently checked other students’ essays and even one of my own, all of which were at least 95% likely to be "authentic." The detector's results aligned with my instincts—this tool appears to be effective!

So, why isn’t this information more widely known?

There could be several reasons OpenAI itself hasn't promoted the Hugging Face GPT-2 Output Detector. Perhaps it is less dependable than my limited testing suggested. Recent news indicates that OpenAI is working on a "watermark" feature as a more dependable security measure. This involves exploring methods to encrypt the texts generated by their language models so that the arrangement of words and punctuation can reveal the text's actual origin. I find this concept intriguing, reminiscent of cloak-and-dagger espionage, yet my experience as an educator leads me to believe that resourceful students—and sometimes hackers—will discover ways around it. Soon, we may find ourselves entrenched in a race between those who encrypt and those who decrypt.

Regardless, I chose to become an educator, not a law enforcement officer. While I generally trust the integrity of most of my students, the emergence of this technology compels me to reconsider my approach to teaching writing.

The first step in my updated strategy will be to adapt. This technology will undoubtedly continue to advance—I expect that soon we will be able to instruct our chatbots to generate essays on any topic in our unique style by providing samples of our previous writing. Therefore, I see part of my role moving forward as staying informed about innovations in AI, much like art instructors have adapted to digital tools over the years. I have always viewed myself primarily as a teacher of art, albeit in the medium of language.

The second aspect of my new strategy involves reconnecting with the reasons that drew me to this profession and engaging in more traditional methods. For those interested in specific teaching techniques, follow me on Medium where I will delve deeper into these topics in future articles.

As for my student, given that my institution lacks an official AI policy, and considering this student was already on track for a C, I ultimately let the matter slide, particularly as it was the beginning of the holiday break. However, I did send them an email inquiring if they had received any AI assistance while writing the paper. Three days have passed, and I have yet to hear back from either the student or their AI helper.

PS — While student dishonesty is a concern, it ranks relatively low on my list of worries regarding this technology. For further exploration, check out my previous articles:

• Is ChatGPT Good for the World: And Who Will We Hold Accountable If It’s Not?
• The Writing Calculator
• My follow-up article: Course Corrections: How to Be an Even Better Writing Teacher in the Age of ChatGPT

If you found this piece insightful, please consider clapping, commenting, sharing, or buying the author a coffee.

Chapter 2: Embracing Technology and Tradition

As the landscape of education shifts with new technologies, teachers must adapt to remain effective. This evolution can be challenging but offers opportunities for growth and innovation.

In the first video, "How To Paraphrase Using AI Without Getting Detected," viewers will learn techniques for using AI tools while maintaining originality and avoiding plagiarism.

The second video, "How to use ChatGPT without plagiarism and AI detection | Tricks revealed," provides valuable insights on utilizing AI responsibly in academic settings.