Recently, there has been discussion that Clemson might join the growing ranks of universities encouraging their students to outsource their thinking to computers. If the University inked a deal with OpenAI that would supply students and faculty with ChatGPT subscriptions, it would tacitly signal its endorsement of large language models in an academic context.
This investment would do a disservice to the University’s students and exemplifies its disregard for actual productive learning. Large language models have no place in most classrooms, and if they must be integrated, they should be incorporated with far stricter guidelines and far more nuanced implementation than Clemson has planned.
The point of a university, at the most fundamental level, is to teach its students how to think. The end goal of most college degree programs is to produce a learner who not only possesses a certain level of subject knowledge but is also able to apply critical thinking skills to put that knowledge into practice. ChatGPT, even in a largely assistive context, detracts from this mission.
When a learner uses a tool like ChatGPT, they are not exercising the mental faculties that would typically perform the task they’ve delegated to the LLM. When you use ChatGPT to generate ideas for your essay, for example, you are eroding your ability to formulate your own ideas.
A recent MIT study found that people who regularly use ChatGPT not only have less developed critical thinking skills, but also become worse writers over time. In a time when literacy rates are falling dramatically, it’s irresponsible of a supposed high seminary of learning to endorse the use of a technology that limits users’ ability to learn.
In addition to the neurological implications of LLM use, they also have a significant impact on the environment. The data centers that power ChatGPT and other LLMs use massive quantities of water and electricity, which both deplete the supply of valuable resources and significantly raise energy costs for people in communities around the data centers.
LLMs and other AI and machine learning technologies do have positive use cases, particularly for data analysis and other scientific applications, but those are very specific uses. The majority of queries that students direct to ChatGPT, though, are not related to data analysis, but instead to generating ideas and summarizing texts.
If Clemson is to adopt ChatGPT and other AI technologies with the fervor that it currently seems to be, the University should specifically and carefully outline appropriate use contexts for these tools, and should discourage or outright forbid their use in reading and essay-writing contexts.
It does students a disservice to have a computer read or write a text for them. We’re in school to learn how to think and how to write. We read texts, especially in the humanities, to understand other people’s perspectives, experiences and knowledge. It’s pointless to outsource that very human endeavor to a computer.
Thomas Merzlak is a junior world cinema major from Florence, South Carolina. Thomas can be reached at [email protected].
