AI Technologies
Anit Vallurupalli
August 23, 2024
Our Education System shouldn't be waging war with AI Tools.
From reintroducing more handwritten assignments, to restricting access to LLM’s on school networks, However, banning LLM’s outright does more harm than good. Teachers should let students use LLMs the same way that they let students use Wikipedia or Google in order to get a good background, before using articles and databases such as Jstor in order to get our statistics and the bulk of our information. This allows students to feel like they are still being efficient with their time compared to using Wikipedia, while also giving the teacher the peace of knowing that it was their own student's work. LLMs can help students with their writing by providing a writing aid in order to spark ideas and to act as a second quality control element There are several reasons why teachers should allow students to use LLMs: although it teaches students that they are not always accurate, it also shows them they should use them like Google or Wikipedia, and still allows them to save time.
The problem with teachers allowing kids to use LLMs is the fact that they will use the LLM for the entire assignment when the real point of these assignments is to develop “ethical values of honesty, creativity, critical thinking skills, and even memory retention" (Gordon)They are not truly learning about the topic when asking ChatGPT or Bard to write it for them, thus wasting everyone's time by asking it to be graded, because they are getting a grade for something for a skill they didn't even acquire or for a topic they don't even know. However, teachers can do several things to make students maintain the use of LLMs as a writing aid and not a text generator for their assignments. Teachers can allow students to use LLM's as a writing aid in order to spark ideas or brainstorm new concepts.
People can convince ChatGPT to do anything because it focuses on user data. As an example, a man named Noble Ackerson searched himself up on chatGPT, and it “insisted I was dead, doubled down when I pushed back, and created a whole new persona.” LLMs are just unreliable data storage spaces, and companies should be more careful when using them (Ackerson). He was able to convince it that he was alive after an hour of what he called “prompt coaxing”, which is telling an LLM that the information is factually inaccurate and telling it another thing until it changes the information present. This shows us that it cannot do overly complicated searches and will stick factually inaccurate information in your face and demand that it be written. However, it is still able to answer your general questions that are slightly more specific than your standard Wikipedia or Google search. So ChatGPT presents itself as a tool for surface level information-gathering, such as Google or Wikipedia, but when we want to know about an in-depth topic, we should turn to scholarly sources that are peer-reviewed.
We already allow students to use Grammarly, and even other students can help edit their essays. However, crucial features are locked behind a paywall, and even if other students can edit someone’s essay, there is no guarantee they missed a run-on sentence or crucial structural errors thhat could ruin the essay. To test this theory, Scribblr, a proofreading service, tested out 10 of the most popular AI grammar checkers. Their AI proofreader and QuillBot both got 20/20 grammar mistakes correct, whereas Grammarly got 11/20 (Caulfield). We have multiple other websites that do this as well, but some teachers don’t know what they are and are scared to let their students use things that will write for them when they should trust them more in order to ensure that they are getting good quality work.
While some teachers have banned LLM usage, this might not be the optimal solution. Instead, they should use a balanced approach of letting students use it akin to Wikipedia and Google. It provides students with an idea generator and allows them to enhance the quality of their writing by using proper grammar and helping students overcome writer's block, as well as providing an additional quality control element. It also teaches students the do’s and don’ts when it comes to AI, as well as teaching them about the limitations and why doing your own work is important. It enhances the learning experiences and prepares teachers for a tech-driven future where AI is an integral part of various professions. It is crucial for educators to reconsider their stances on LLMs, adopting a more inclusive approach that allows students to benefit from them while educating them on ethical usage. LLMs in the education sector offer a unique opportunity to blend traditional methods with innovative ones, and in the process will enrich the learning journey for all parties involved.
Works Cited
Ackerson, Noble. “GPT Is an Unreliable Information Store.” Medium, 27 Feb. 2023, towardsdatascience.com/chatgpt-insists-i-am-dead-and-the-problem-with-language-models-db5a36c22f11.
Burrows, Spencer . “Three Steps to Prevent ChatGPT Misuse.” Education Week, 7 Mar. 2023, www.edweek.org/technology/opinion-three-steps-to-prevent-chatgpt-misuse/2023/03.
Caulfield, Jack. “10 Best Free Grammar Checkers | Tested & Reviewed.” Scribbr, 6 Oct. 2022, www.scribbr.com/language-rules/best-grammar-checker/. Accessed 23 Sept. 2023.
Gordon, Cindy. “How Are Educators Reacting to Chat GPT?” Forbes, 30 Apr. 2023, www.forbes.com/sites/cindygordon/2023/04/30/how-are-educators-reacting-to-chat-gpt/.
Hernandez, Michael. “AI-Resistant Assignments? Show Student Thinking and Promote Better Writing with UChicago-Supported Tools | Academic Technology Solutions.” Academictech.uchicago.edu, 8 Feb. 2023, academictech.uchicago.edu/2023/02/08/ai-resistant-assignments-show-student-thinking-and-promote-better-writing-with-uchicago-supported-tools/. Accessed 23 Sept. 2023.
