ChatGPT: Ethics in Academics

ChatGPT is an artificial intelligence (AI) language processor, or chatbot, that was created by OpenAI in December 2022. It can interact with users in a conversational way, with features such as the ability to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests. ChatGPT is an improved version from a model in the GPT-3.5 series by becoming a safer and more useful AI system. There has been a substantial reduction in harmful and untruthful outputs by the use of reinforcement learning from human feedback. However, because of ChatGPT’s ability to respond to users accurately, there comes a new problem: The use of the AI for school work. When ChatGPT first launched, I overheard multiple students here at Georgetown saying they used the system to write final papers for them and received a high grade for it. Nonetheless, although ChatGPT is ethical to help students understand a concept, it is unethical to claim the product of ChatGPT as their own work.

While we think it is ethical to use ChatGPT to help with understanding concepts that are taught in class, it is unethical to rely on ChatGPT to directly submit coursework assignments. For example, a student may ask ChatGPT questions like they would to a TA to help them understand key concepts, but the application of the concept to a homework question ought to be completed without the aid of ChatGPT. Beyond the obvious reason that an assignment explicitly states to not use external aid, students wouldn’t be able to properly develop their independent and creative thinking if they were to rely on ChatGPT. Thus, using ChatGPT deprives students of the opportunity to develop critical thinking and communication skills. Furthermore, when the professor gives feedback on the assignment, they’d be commenting on how to improve ChatGPT’s work, and the student would lose the opportunity to improve their analytical abilities.

Some may conclude that since courses aim to prepare students to enter the workforce and that since ChatGPT is allowed in the workforce, students should be allowed to use ChatGPT as a tool to complete coursework. Researchers are already using ChatGPT to generate code to help them analyze data. Why not use this tool if it is available? However, if students were to use ChatGPT to directly complete an assignment without fully understanding their submission, they would be making a risky assumption that the output of ChatGPT is correct. Making this assumption in the workplace could have disastrous consequences if one were to turn in work they didn’t understand and that turned out to be wrong.

We also argue against the use of ChatGPT as a means to boost one’s grades on idealistic grounds. The noble pursuit of truth and goodness does not rely on shortcuts. The satisfaction of turning in your original work comes only after spending hours to understand a concept, to revise a paper or to solve a problem. Thus, ChatGPT does not further a student’s intellectual development if they were to use it solely as a means to click that submit button on Canvas.

In addition to simply submitting assignments, there is also another issue of citing sources when using ChatGPT. The AI is trained on vast amounts of information from the internet including websites, books, and news articles. However, when it responds to the questions sent by the users, it does not cite where it compiled its information from. Although it is difficult, almost impossible, to cite all the sources that ChatGPT got its data from because of the possible vastness, without citations in academic papers, the work will not be credible. Citations make the researchers and scholars credible by acknowledging their work. The academic community is interdependent; without relying on other people’s works, researchers cannot have something to work off of to expand on a point or refute it. As students, we research topics based on the classes we take and the majors we choose. While not everyone moves on to further education past the undergraduate level, it is good practice to cite sources when possible to recognize the others’ work and contribution to society. Without this opportunity, students may graduate without the skills necessary to be part of the interdependent society.

Ultimately, there are limits to the use of ChatGPT in the academic setting on ethical grounds. Relying on ChatGPT to solely submit a coursework assignment deprives students of the opportunity to develop their creativity and form their opinions among many other skills. Additional ethical concerns include the inability to cite credible sources, as well as factually inaccurate outputs. However, we acknowledge that natural language processors are still in their early stages, and developing at a rapid pace. There is potential for education systems to leverage AI to accelerate students’ intellectual growth while addressing issues of citation and cheating.