LLMs in Education: Forged from Years of Educational Challenges

AI developers have rapidly integrated their products into society, including the education sector. This has made us, the educators, feel overwhelmed and threatened, and rightfully so. Writing assistants powered by LLMs pose a serious problem, or at least appear to. One common worry among teachers is students cheating. To address this, they are scrambling to develop alternative assessment methods because essays risk becoming obsolete. However, beyond cheating, many educators are worried about students’ learning outcomes, the harm to education, and the existential risk to composition teachers. Consequently, they advocate for AI detectors or watermarking AI-produced text as deterrents.

Like most other inventions, LLMs have both negative and positive consequences. For example, LLMs might particularly benefit marginalized populations, such as students with weak English composition skills who feel confident and less anxious about their writing. They can ask ChatGPT for help with checking grammar, editing, improving sentence structure, and providing feedback. This support is crucial for international, immigrant, and non-native English-speaking students and all other marginalized students without extra tutoring or teacher oversight. These students often face challenges in secondary and post-secondary English courses. It is no secret that many students, native speakers or not, graduate high school with inadequate composition skills. Some professors report that even graduate students sometimes lack the expected skills.

To help make my point, let me share a brief personal story.

Not long ago, I discovered something about my time in middle school. As a child of immigrants, I struggled in some classes, especially English. To help me, the school decided to pull me out of my regular English class once or twice a week and place me in a remedial class. Now, I remember attending this class but only recently understood why. Neither I nor my parents recall anyone asking for permission or discussing this move. I was just set aside. I felt the regular English teacher was annoyed by students like me and only paid attention to a few ‘star’ students. Until recently, I had no idea the other class was remedial. I realized it during a conversation with a friend, a native sixth-generation Oklahoman who is now an excellent teacher and was also placed in a similar class. Ironically, she felt the same about the experience. We both remember the remedial class being pointless, though her experience was slightly better. I recall doing virtually nothing in that class. The teacher would leave five minutes in, and a teacher’s aide would play games. For a semester, maybe longer, we spent every class doing very little.

Many might argue that our experience is an exception and not the norm. Maybe, but declining reading and writing scores among high school graduates are not anomalies. Many colleges offer 0-credit hour courses to address ill-prepared students, though I doubt these classes are helpful. Almost everything I know about writing, for better or worse, is self-taught. The point is, why shouldn’t students ignored by teachers in high school and early college use LLMs to level the playing field? LLMs can be an equalizer. Students no longer need to stress about bad grades because they can’t communicate ideas as well as those not left behind by the system.

Some might say, “Let’s fix the system instead of relying on AI.” However, students struggling today lack the luxury of waiting for a system that is perpetually under construction. They have to compete now and have every right to use any tool that helps level the playing field.

Why am I telling you all this? The threat LLMs pose to composition instructors and others, such as philosophy faculty has been on the horizon for quite a while. We’ve treated students in these courses like objects on a conveyor belt and the courses like a required product—a product every student must take, which brought a sense of job security and, unfortunately, cultivated complacency. To be clear, many teachers reject this attitude and strive to do their best, but many have not. Moreover, my grievances apply to all students who struggle in these courses, not just immigrants and non-native speakers, but people like my friend. I am talking about all the students who are ignored by the system because the system feels that helping these students is not worth their time or pay (of course, composition teachers should be paid better).

Thus, the issue extends beyond teachers’ attitudes and into a societal disregard for skills like reading and writing. It’s common to hear composition teachers lament, “These students can’t write,” “They seem entirely uninterested,” and “It’s like pulling teeth,” and most of the time, they’re referring to native speakers. A quick Google search reveals several results, from formal studies and opinion pieces to Reddit forums, discussing the terrible writing skills of high school seniors, first-year college students, and even graduate students.

LLMs have exposed deep wounds that have metastasized over many years: the lackluster attitude of many teachers, the demeaning of liberal arts education, the creation of an environment where composition courses are only deemed worthy of being taught by graduate students or part-time instructors, and students frustrated with the system. Consider this quote from a sixteen-year-old article in Inside Higher Ed: “If writing is so important that virtually every student at nearly every college and university must take at least one composition course (usually two), why aren’t more professors of English teaching it?”. Indeed, why? Claiming that composition is crucial, while departments and institutions are only willing to spare overworked and underpaid graduate students and adjunct instructors to teach it, signals triviality. If we want students to care about writing, let’s value the people and courses teaching it.

I will end with these final thoughts. When there is a vulnerability in a system, some innovate to help solve it, and others exploit it for personal gain. And which one are most AI developers? I do not yet know. Unquestionably, LLMs create ethical and pedagogical problems for composition and other writing courses, but for many students, they have provided a solution. It solves a problem for many students by helping them achieve adequate grades. However, some might argue that the real issue is that students pass through classes without mastering essential reading and writing skills. Before the advent of commercial LLMs, some institutions, instructors, and parts of society ignored struggling students in acquiring these critical skills, yet expected academic performance.

Nonetheless, after struggling through high school, these students often arrive at college still viewed as burdens—unworthy of the attention of tenured, experienced professors and undeserving of the instructor’s time. In such an environment, it is unsurprising that LLMs are flourishing.

Writing excellent essays may indeed be important for all students, and I am sympathetic to this ideal (yes, it is an ideal). However, should these students care now when LLMs can help them achieve the same high grades that previously denied them opportunities similar to those of their peers? It seems that grades were what mattered to many of their teachers, not whether the students learned these skills. Since many in charge did not prioritize ensuring these students acquired proper comprehension skills before LLMs, their sudden concern might ring hollow.

Lastly, while it might seem like I’m placing all the blame on the instructors, that’s not my intention. Many instructors do their best with the little resources they are given. However, the attitudes of many others towards struggling students and the course have clearly negatively impacted the attitudes of students and other stakeholders. Dissatisfaction with the education system is increasing. For instance, a Pew Research poll from April (2024) found that 51% of Americans believe K-12 education is trending in the wrong direction, and 57% do not trust public school teachers (much of this is politically motivated). Unless crucial aspects change on several levels, the future looks bleak for our profession.

Syed Abumusab

Website

Syed AbuMusab is a Postdoctoral Associate at the University of Bologna and Yale’s Digital Ethics Center, where he specializes in the social aspects of artificial intelligence and other emerging technologies. His research focuses on AI’s social ontology, agential capacities, social influence, and ethics. For example, how do generative AI systems like Large Language Models impact core social institutions and practices such as education, the judicial system, and interpersonal relationships? Additionally, his work explores the philosophy of mind and computation, examining how these areas intersect with emerging technologies.