Artificial intelligence (AI) can supposedly solve any grammatical query for us. If it can do that, does it make learning grammar obsolete? What should English instructors teach their students, then?

As increasingly advanced AI technology starts to challenge writing, we need to reevaluate what it means to be able to write. How can students wield AI technology to their advantage in order to succeed in the digital age? Where does AI trump human skill, and where does it fall short?

Educators have widely addressed generative AI such as ChatGPT, but one platform tends to fly under the radar in high school classrooms: Grammarly.

What is Grammarly?

Grammarly is a digital writing application that uses AI to analyze user-submitted text. It provides suggestions to improve grammar, style, clarity and more. The platform works by underlining identified errors with red and blue lines, which users can accept or decline with a click. It also has a lesser-known generative AI feature akin to ChatGPT, which can rewrite a given sentence to satisfy a goal.

To demonstrate Grammarly’s primary function, I provided it with the following grammatically flawed sentence:

Mike, climbing the tree saw few apples that were very red.

Grammarly’s suggestions reworked the entire sentence. It now read:

Mike climbed the tree and saw a few very red apples.

Grammarly makes the writing much better, right? The students at my all-girls private high school in New York City certainly think so. They often use the platform as a sort of crutch to fall back upon. The prevailing attitude is, “My essay is bad, but that’s okay — I’ll just put it through Grammarly.” They laud the technology as “a lifesaver.” But is it really?

Investigating Grammarly in high school: test and results

I wanted to pinpoint how Grammarly affects student performance, as well as identify when it proves useful or backfires. To this end, I conducted a proper test with high schoolers, analyzed the results, recorded statements from teachers and used the platform’s chatbot to make additional observations. I considered all of this information and reached a conclusion: Students should use Grammarly only to learn the mechanics of writing, then rely on their own skills once they’ve grown confident in their grammatical knowledge.

I recruited an English teacher and eight of my fellow sophomores for this experiment. The study’s purpose was to assess in what ways Grammarly may affect how students write for English class.

I started by crafting a sufficiently thought-provoking prompt. After all, students tend to make more grammatical errors when attempting to relay complex ideas. I settled on the following instruction: “Please write 300–500 words on how the main character of your favorite book, show, or movie relates to an aspect of your identity and represents a broader lesson for individuals or society.”

I had four of the students edit their writing with Grammarly. To simulate how students use the platform in practice, I did not give overly specific instructions but told each student to utilize the tool as they saw fit.

To the other four students, I stressed that they should avoid all outside assistance. None of the students knew the full picture of the experiment. The participants could not look at the prompt until they were ready to begin writing.

I allotted each student a single 30-minute session to complete their paragraph. Each response was graded by a high school English teacher based on the student’s style, diction, syntax, clarity and grammar, irrespective of their chosen subject matter. The teacher also attempted to identify the works that had gone through Grammarly.

Overall, there was no distinct correlation between Grammarly usage and higher scores. The highest-scoring piece, which received 96%, did not use the platform. Of the four highest-scoring and four lowest-scoring pieces, two of each used Grammarly.

When choosing the students to participate in other study, I enlisted ones with different grades in English class. Four of them averaged grades between 94% and 100%; the other four between 90% and 93%. In a selective school like this one, such a difference between these grade-average ranges meant a significant distinction in writing abilities. Two students from each group used Grammarly on their pieces.

The teacher correctly identified Grammarly usage, or lack thereof, in six out of the eight pieces.

Five of the students received scores in this exercise that were lower than their average scores; one received a higher score, and two received average scores. At first glance, Grammarly did not seem to affect whether students surpassed their own records. Of the five students who performed worse, three used Grammarly. Of the two students whose grades remained consistent, one used Grammarly. The sole student who scored better than their average grade used Grammarly.

Upon deeper analysis, Grammarly maintained or improved the grades of lower-performing students while negatively impacting higher-performing ones. Of the two students in the lower-performing group that used Grammarly, one scored better than their average grade (94%-) and one scored the same (90%). The two students in the lower-performing group that did not use Grammarly scored below their average grades, scoring an 84% and an 89%.

Of the two students in the higher-performing group that used Grammarly, both scored much lower than their average grades (85%). The students in the higher-performing group that did not use Grammarly scored higher than their average grades, scoring a 96% and an 88%.

This result seems relatively clear-cut. Still, it must be noted that the writing skills of same-grade students are not necessarily identical, which could skew how Grammarly changes students’ grades, from their average grade to their exercise grade.

While it makes sense that Grammarly should improve weaker students’ writing, it is striking how Grammarly is detrimental to higher-ability students. Maybe Grammarly is not as efficacious as it claims to be. On the paragraph of one higher-performing student who used Grammarly, the English teacher’s critiques concerned “usage rather than things that [were] grammatically ‘wrong.’” The teacher further said that “the writing [was] awkward, with syntactic issues.”

Similarly, for the other high-performing student who used Grammarly, the teacher stated: “I marked a few ‘errors’ that are really more about usage than exact rules. Yet I found this [essay] hard to follow overall. I would not be surprised if it had been through Grammarly.”

He told me these were the two pieces he was most certain used Grammarly. He was correct.

Grammarly vs human artistry

The cardinal purpose of grammar is to illuminate and elevate the meaning of writing. Grammarly was able to improve the technicality of the higher-ability students’ writing, but it concurrently diminished its cohesivity and impact. Perhaps there is a deeper flaw in the platform, one regarding its artificiality in contrast to the innate humanity of writing.

Regarding the highest-scoring response, which was written by an A student who did not use Grammarly, the English teacher remarked upon the vividness of the student’s writing style: “The writing here is strong, in a way that transcends grammar. The writer purposefully uses polysyndeton and asyndeton (repeating or excluding conjunctions) even though they are technically grammatically ‘incorrect.’” Had the student used Grammarly, the platform would have “fixed” those errors, thereby damaging their writing quality.

Writing is more than a set of hard rules — it is a craft. Proficiency in this field requires a deft knowledge of how to both follow and break grammatical rules. Grammarly’s unyielding technicality is incompatible with the holism of writing.

“Grammarly is just another machine to flatten our multidimensionality,” Dr. Paul Phelps, another high school English teacher, told me. By feeding it a passage written by English novelist Jane Austen, Phelps further highlighted what my study revealed about the platform. “I can tell you this,” he said, “Grammarly thinks that Jane Austen is a terrible writer. Grammarly wants to change everything about Jane Austen.”

Even generative AI does not make Grammarly creative

Grammarly is no substitute for human ingenuity — not only in regards to grammar, but also in producing “original” writing. I learned this when I tried out the platform’s generative AI chatbot feature. I provided the same Grammarly-revised sentence from earlier — “Mike climbed the tree and saw a few very red apples” — and asked the platform to change it in various ways. I instructed it to make the sentence persuasive. Grammarly gave:

Mike skillfully climbed the tree and was thrilled to discover several irresistible, vividly red apples.

Next, I told it to make the sentence funny and creative. Grammarly returned:

Mike channeled his inner monkey and ascended the tree like a pro, only to be greeted by a bunch of apples that were more tempting than a chocolate fountain at a dessert buffet.

Then I told it to make the sentence so sad that it would move a reader to tears. Grammarly spat back:

Mike, struggling with a heavy heart, ascended the tree, his mind weighed down by the burden of loneliness and regret. At the top, he was met with a cluster of apples, their allure mocking his unfulfilled desires and lost opportunities, leaving him feeling more isolated than ever before.

Finally, to really see what the platform could do, I told it to make the sentence sound like it was written by former US President Donald Trump. Grammarly concluded:

Nobody climbs trees like Mike. The best tree climber, believe me. And let me tell you, he saw some incredible, very red apples. I mean, really tremendous apples. The reddest apples you’ve ever seen, folks.

Grammarly’s writing abilities are mediocre, to put it bluntly. Its generated sentences are bland and banal, and its metaphors are overwritten and drawn out. Its writing is dull, laughable and clunky. Its descriptions sound phony and generic. However, while the platform is certainly no match for the great wordsmiths of literature, its writing is functionable. It is clear and sensible.

Grammarly can be a handy tool for practical purposes. Writing rarely needs to be revolutionary outside of an artistic setting. Most people are not aspiring authors, nor do they want to produce groundbreaking prose. However, most people need to send clear emails, memos and letters, tasks for which the platform would be a boon.

Grammarly’s impact on student learning

Some teachers have argued that using Grammarly deprives students of learning. Phelps described the consequences of Grammarly usage as “a sort of weakening of confidence.” He argued that while the platform is “the path of least resistance, it’s also the path that will produce the least amount of learning.”

Yet other data suggest Grammarly may help students improve their writing. In 2022, Negeri 1 Ujungbatu, a high school in Indonesia’s Riau province, performed an experimental study in which students took separate writing tests before and after using Grammarly on other work. This study found that the platform yielded a significant improvement in student scores. Likewise, in a 2011 survey conducted by Grammarly, Inc., 70% of the 392 student respondents reported that the platform had increased their confidence in their writing abilities.

To get a straight answer on this matter, I turned to Grammarly itself. I asked the AI chatbot whether high school students’ usage of Grammarly comes at the cost of their learning. It replied with an answer that, while true, overlooked one critical detail:

Using Grammarly does not necessarily come at the cost of learning. While Grammarly can help identify and correct mistakes, students still have the opportunity to learn from these corrections. They are able to see explanations for the suggested changes and can gain a deeper understanding of grammar and writing conventions.

Though this is a valid statement, Grammarly’s simplicity discourages users from taking advantage of its educational function. Users can click a button to view the justification for its suggested edits, but they are not required to read the explanation before accepting it. So, many won’t. Likely, Grammarly did not account for this easy bypass in its answer because its AI is programmed to uphold its image as integritous and meritful for classroom use. 

“The issue is that [Grammarly] is so automatic that you can just hit check,” a high school senior told me. “I’m not sure if someone would actually pay that much attention to what it’s saying.”

Phelps concurred. “Nothing Grammarly does is hard to learn. Nothing. It is not difficult to learn whether or not you need a comma between two independent clauses. But if you’re not invested in that learning or asking about it, you’ll never learn it.”

However, even without reviewing the reasons behind the suggestions, I believe students can learn from Grammarly. As they click on underlined words, accept changes and see the red color vanish before their eyes, their brains can develop a subliminal aversion to these colored underlines. They will then unconsciously work to write in a way that avoids triggering these error marks.

Properly wielding Grammarly

Does learning grammar even matter if Grammarly will always be available? I would argue that, yes, students do need to learn the laws of grammar. Doing so enables students to discern precisely how and when to use AI tools. For their own benefit, they should use Grammarly as an assist to learn writing mechanics; once they are more self-assured, they must detach from it and take the wheel themselves.

We must teach students when they can stop using Grammarly: when they inevitably transcend its abilities. As my data has illustrated thus far, the platform has its limits, and human thinking can surpass them.

Imparting this technological savvy upon students equips them for their professional lives. Their competition will undoubtedly have access to Grammarly. While this platform is not necessarily the best-suited for a high school setting in which students must complete tests with pencil and paper, there will be very few instances in adult life where Grammarly is inaccessible.

Ultimately, we must remember that Grammarly is simply a machine, one that we can employ but should not rely upon. As Phelps told me, “None of these [AI] devices are independent of insidiousness. If they are not used carefully, they can be as harmful as helpful. Unfortunately, sometimes the harm and the help can be the same thing.”

With that said, wholly rejecting useful technology would be like driving a horse-drawn carriage instead of a car just so you can develop that skill. The best way for us to get ahead is to use AI like Grammarly as a supplement to, rather than replacement for, our own unparalleled human prowess. This is what students need to learn.

[Lee Thompson-Kolar — not Grammarly — edited this piece.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.