Article: Teachers Beware — How Students Can Cheat with AI

Photo Source: Pexels

When it comes to expository essays that so many kids dread writing, students just got a big boost from chatGPT. On November 30, 2022, OpenAI, an artificial intelligence company based in San Francisco, released the latest version of its large language model called chatGPT. GPT stands for “Generative Pre-trained Transformer,” and chatGPT represents the newest version of OpenAI’s groundbreaking GPT models. Large language models generate human-like answers to human questions.

A transformer-based program such as chatGPT uses a type of artificial intelligence called deep learning. Deep learning works somewhat like the way people learn through repetition and practice. In other words, a transformer learns from text by reading through millions of documents, periodically hiding words from itself, and trying to guess what those words should be. When chatGPT guesses a word correctly, it strengthens itself. Just as students do fill-in-the-blank tests to assess their knowledge of a subject, chatGPT challenges itself with fill-in-the-blank tests while reading through the entire internet. Over the course of billions of tests, chatGPT better understands and replies to human language, having learned which words make the most sense together.

Following its release before Christmas, ChatGPT elicited immediate and vibrant responses from the press. The New York Post wrote on December 26, 2022, about a professor who caught one of his students using chatGPT to answer a take-home essay test for a philosophy class. (nypost.com) The student admitted to the charge and was kicked out of school. The professor admitted the difficulty in proving that chatGPT wrote the essay. Others do not see chatGPT as an enemy of teaching. Rather, Business Insider reports chatGPT as a potentially helpful tool. (businessinsider.com) A professor from Penn State University considers chatGPT a valuable aid for students to refine their work, but chatGPT will not help with original writing or complex argument construction.

Anyone can try chatGPT on the OpenAI website (chat.openai.com). I tested it with a question below.

My question: “How did the paintings of Jean-Michel Basquiat influence the art world?”

ChatGPT response: “Jean-Michel Basquiat’s paintings, which combined elements of graffiti, primitive art, and abstract expressionism, had a significant impact on the art world in the 1980s. He emerged as a leading figure in the Neo-Expressionist movement and his work was recognized for its raw energy, social commentary, and fusion of text and image. Basquiat’s paintings often explored themes of race, class, and power, and his unique style and use of iconography resonated with a wide range of audiences. His work helped to break down traditional barriers in the art world and paved the way for other artists of color to gain recognition. Additionally, his meteoric rise to fame and untimely death at the age of 27, cemented him as an art world legend and continue to influence artist today.” (January 17, 2023)

ChatGPT wrote the above paragraph in about 5 seconds, and it has some interesting details and covers some general facets of Basquiat’s life. Interestingly, Grammarly, the AI grammar checker, had many issues with chatGPT’s grammar and use of passive voice. I ran the question a second time on chatGPT and got a slightly different answer. A different answer makes sense because the model writes from scratch each time, which makes plagiarism checks more difficult.

OpenAI released chatGPT in the late fall of 2022, and the press immediately took various positions regarding the implications of this powerful large language model. Large language models learn by reading millions of pages of text, and during the reading, the model removes words from sentences and tries to guess the right word. Every time it guesses the correct word, the model grows more robust and more accurate. In the case of chatGPT, the model has a very conversational tone. Some see the rise of large language models as a dangerous tool to help a student cheat. Whereas others see it as a new tool to help students refine their work. I found the output of chatGPT very interesting due to its command of language and general, thoughtful coverage of my question. We shall see as large language models grow whether we use these as tools for greater creativity or to avoid the hard work of learning to write and communicate through the written word.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.