To balance AI in academia, the education system requires restructuring
To support student journalism and the content you love, become a member of The Daily Orange today.
While the increasing complexity of artificial intelligence is awe-inspiring, there are questions on what it will mean for students and the education system.
OpenAI, an artificial intelligence research lab, has been making headlines recently for its new language modeling system that is defining the current complexities of AI. The newest iteration, GPT-4, launched on March 13 and is accessible through paid subscription. The free version of ChatGPT still exists, with 13 million individual active users visiting the site per day as of January 2023.
Generative Pre-trained Transformer describes the technology’s ability to generate text when prompted based on extensive information the AI has been trained to learn. The newest version of the machine, GPT-4, can process both text and image data, and is much less likely to provide incorrect information than ChatGPT — meaning it will only adapt as each software comes out.
The way these models work is that someone can provide the program a topic to write about and the AI will scour the internet for information. Then the writing is generated by the AI determining which keywords are often used when discussing the given prompt and stitching them together into a written piece.
Beyond harnessing the AI for written assignments, GPT-4 was recently proven to rise to the occasion on standardized testing as well. OpenAI claimed GPT-4 can score a 1410 on the SAT and pass Advanced Placement exams and bar exams alike.
Standardized testing was created to judge human education. But if a person’s educational level can be equaled by a bot, then are these tests proving someone’s learned skills or rote memorization?
In general, these advances are troubling, as they upend our traditional understanding of education. Many are left to wonder if standardized testing is now obsolete because a bot used to generate predictive text can easily excel on them. As student utilization of AI rises, written assignments could end up being the work of AI, rather than organic material from a student.
As the site gains traction, one of the more contentious debates surrounding this new AI is its ramifications regarding academia. Due to its ability to generate predictive text, many students have been getting away with entering prompts from their assignments and letting the AI write the assignment for them.
The student opinion is varied on the subject of AI. Some believe completing assignments with the AI is ethical. Others take a more pessimistic outlook, with ChatGPT making them wonder why they should put effort into schoolwork when they know a chatbot can peform better than they can.
Stephanie Zaso | Digital Design Director
It’s easy to understand the nature of this hopelessness. The school system, as it stands, focuses less on genuine comprehensive understanding from students and more on the regurgitation of information. Assignments such as multiple choice tests are more conducive to assessing pattern recognition and guesswork than any real knowledge. There is no great effort in assuring that any information has been retained.
A machine’s ability to perform well on a test that is meant to evaluate students’ knowledge and progress calls into question the effectiveness of our current education and testing methods, like the SAT. A high score may appear to be a marker of collegiate level ability, but the performance of AI forces us to reexamine what abilities are truly being marked and asked of students.
The threat is more imminent regarding written assignments. While a human’s opinion, grounded in lived experience, can never be effectively replicated by a bot, most college-level essays are asking for arguments on the basis of research. That research may be part of ChatGPT’s training data, making it easy to recall for an assignment.
This is especially troubling in the humanities field, where a degree is based highly on your performance on essays. If AI gets to the point of writing works indistinguishable from a student’s writing, the field as a whole could become one dominated by bots rather than people, similarly for other writing intensive fields.
If a student can produce a paper that can’t be identified as the work of a bot, then it is unclear how college professors and teaching assistants, already tasked with grading many of these papers, will know which are authentic.
Students must also find their footing when it comes to AI. Complete dependence will never amount to the lived education of students. For all its full exposure to everything the internet has to offer, an AI can never replicate the experience of learning under a professional in the field, nor the personal voice we each individually bring.
Due to the implications that AI has on both standardized testing and the traditional written assignments, it is clear that a restructure of the education system as we currently know it is the only way to effectively find the balance and collaboration of AI in an academic field.
In its current state, the education system demands a rigid, egoless approach. An answer is either correct or incorrect, what with standardized tests made specifically to be scored by machines. There is no place for personal pronouns or anecdotes within written assignments. As it stands, the distinct lack of creativity demanded from students results in alienation from the information they are meant to be internalizing.
AI does not represent an antithesis to this pursuit, but rather a tool to be used, albeit sparingly. While AI has the ability to encourage academic dishonesty, it would be short-sighted to see this innovation as a singularly negative thing. Due to the bots’ ability to generate predictive text, AI can inspire ideas branching from the original thinking provided by a user. AI used in this sense can provide groundwork and ideas for those using it instead of being relied upon to complete entire assignments.
The ability of AI, such as GPT, sheds light onto what the education system deems to be academic prowess as it reveals the ease in which computers can replicate the work demanded of students. The school system ultimately cannot affect the progression of AI, but we can control how this progression is responded to.
The most productive way forward is to repurpose how we approach assessing students’ performance while acknowledging that AI is not merely something that can be used as a means to cheat, but also a very powerful tool to work in conjunction with.
Summer Brannan Taylor is a freshman in the College of Arts and Sciences
