“With artificial intelligence, we’re summoning the demon.” It’s safe to say that Elon Musk isn’t the biggest fan of artificial intelligence. His beliefs echo a widely held sentiment in America today: AI is simply too dangerous to take a chance on using it for good. With the release of OpenAI’s ChatGPT, the first publicly accessible artificial intelligence, this concern has finally hit home for many Americans, especially teachers. Will this new tool improve student performance, or will it damage the fragile recovery made since the conclusion of the pandemic? Well… it’s just not that simple.
Since the dawn of time, teachers have strived to make homework assignments as airtight as possible. For centuries, a watchful eye in the classroom was enough to quell cheating, but then came the Internet. Let’s face it; the Internet has made cheating on homework assignments far easier for the average student. Every problem in a standardized textbook is available behind an insignificant paywall or even for free. However, the Internet has also been a blessing for learning. Confused students can look to Youtube for step by step tutorials on any subject under the sun, and instead of only hearing a teacher’s lecture a single time, the Internet has created the capacity for students to view lessons as much as they want, when they want, until they have a complete understanding of the subject.
ChatGPT takes this concept to the next level. In addition to having the potential to explain complex problems, it can also engage in meaningful dialogue for additional questions a student might have, anytime, anywhere. But this amazing capacity is largely limited to learning, not actual assignment completion. Teachers shouldn’t be too worried: cheating with ChatGPT is harder than it looks.
Yes, ChatGPT has the capability to write essays as of now, but its writing abilities at these early stages are extremely limited and cannot be used for quality writing in high-level classes. For example, I took the first paragraph of an opinion piece I wrote for the Indy Star and fed it into ChatGPT, instructing it to “improve and edit this article as if you were a newspaper editor.” Needless to say, it completely missed the sarcasm in the opening lines butchered the intent of the paragraph (here). That’s because although ChatGPT has a great understanding of grammar as a language model, it’s severely lacking in the creativity department, and every answer that it will ever spit out has a basis in something already existing on the internet. As another test, I requested that it “create a plan for a successful investment firm from scratch.” The chatbot replied with a slightly paraphrased business model of the already existing company, Wealthfront, without even bothering to change the company name (here). Is it impressive that a machine can produce that in under ten seconds? Yes, but even so, ChatGPT isn’t very good at creating new ideas; it’s essentially just an intelligent version of Google. As NCHS junior Luke Roan put it, “You’ll probably end up taking more time just to get a workable prompt than you would writing the essay.”
The chatbot has undeniably been popularized by its pliability, the ability of users to ask roundabout questions to make it perform tasks it isn’t actually supposed to do. Tiktok users have showcased this flaw repeatedly in viral videos, and this is likely what concerns teachers and supervisors the most. Ironically, however, it is this weakness that makes ChatGPT so strong. OpenAI publicly released ChatGPT for “fine-tuning;” every user’s chat is analyzed, and when breakthroughs are found, they are sealed very quickly. Eventually, discoveries on ChatGPT will plateau, and it will become merely another resource for searching the internet.
This brings up another important point. ChatGPT’s access to the internet has been cut off as of 2021, and its information is slowly becoming more and more outdated as our knowledge improves. Denying students access to this resource now when it is most helpful is risky; we may never see another publicly accessible AI like this again. Even ChatGPT’s parent company, OpenAI, plans to use the data they are gaining from this experiment to make an improved, paywall-hidden version of the chatbot. Teachers should take advantage of this unique opportunity while they have it, and maybe even try to incorporate exploration of it into their classroom schedules.
ChatGPT will never best real human creativity. Teachers and students shouldn’t be afraid of change they cannot stop. Instead, let’s embrace it and explore the endless possibilities that lie ahead in the realm of artificial intelligence.
