Grammarly: The Good, the Bad, and the AI Component in Our English Language Writing

By Nihan Sanic, University of Illinois Urbana-Champaign

DOI: https://www.doi.org/10.69732/EMVO1008

Founded in 2009, Grammarly started as a basic spell checker and quickly evolved into a full-fledged AI-powered writing assistant. Unlike spell checkers, Grammarly provides real-time feedback in four categories. Correctness underlines grammar, spelling, and punctuation errors, while Clarity focuses on the conciseness of the text and avoiding wordiness. Engagement looks into the word variety to avoid dull phrasing, and Delivery is about the tone, politeness, and formality of the paper (Lytvyn, 2022). Additionally, the tool recently came out with its own version of Generative AI alongside its AI checker. As the tool developed, it became quite mainstream amongst learners of English (along with various similar tools that exist for writing in a variety of world languages). Available as a browser extension, desktop app, mobile app, and even inside Google and Microsoft documents, Grammarly can be as embedded into the writing process as a keyboard. 

The integration process starts innocently enough for these students: a red squiggle here for punctuation, a helpful phrase tweak there. However, for some, before they know it, their paragraphs sound like a courteous bot wrote it, that might as well have the tell-tale sign: “Sure, here you go!”. Therefore, while Grammarly can be an excellent tool for second language learners and native speakers alike, it also raises some red flags through extensive use, such as overreliance and inconsistent writing, especially with the paid Grammarly Pro. The free version of Grammarly can be classified as a solid proofreader for errors when it comes to correctness, and provides some information on conciseness and citations with limited AI generation. The Pro version offers advanced suggestions for writing style and tone enhancement, catching more complex problems from the perspective of academic integrity. It provides tailored feedback for formal writing and professional communications and reports personalized insights and statistics for users as well. However, perhaps most notably, Grammarly Pro has full access to the company’s generative AI with around two thousand monthly prompts, which complicates things as the tool evolves from an editor to a content creator, blurring the lines between writing with an aid and AI-generated writing.

The Good: How Grammarly Helps

As can be expected, Grammarly excels at detecting and correcting grammar and spelling mistakes, making it an efficient proofreader. It catches both basic and advanced mistakes, including subject-verb agreement, misplaced modifiers, spelling issues, and run-on sentences. It is also quite effective at providing suggestions for sentence-level clarity. For example, when the sentence ‘In the event that you are late, please be advised that you should inform the instructor’ is entered, the tool suggests cutting it down to ‘If you’re late, please inform the instructor.’ Additionally, the tool excels at supporting non-native English speakers by suggesting natural-sounding phrasing and correcting idioms for writers. Of course, Grammarly is both a time and grade saver tool for many students. It is useful for quick fixes in emails, reports, and essays, and with generative AI – for better or for worse – these bodies of text can be written with twice the confidence and half the time it would take learners on their own. The tool checks for errors, reducing the constant need for proofreading, which saves time and leads to better grades on the grammar aspect in the rubric due to these automatically suggested changes; perhaps even lessening the Academic Writing Anxiety for learners, depending on their trust in the tool. In short, Grammarly is an excellent tool when used as an assistant alongside personal writing skills.

The Bad: When Grammarly Falls Short

As mentioned, the tool excels when used alongside pre-existing skills; however, concerning lapses start appearing when learners stop critically engaging with the tool. Grammarly is still improving on overcorrections and incorrect suggestions, and sometimes fixes things that aren’t broken, making sentences sound robotic. For instance, when the sentence ‘I feel like he’s not being honest.’ is entered, the tools suggest sentences such as ‘He is not being honest’ or ‘I feel as though he is not being honest.’ which leads to unnecessary formalization or loses the original meaning of the sentence. If students do not carefully consider the suggestions prior to accepting them, they may end up with sentences with unintended meanings, which can also be seen in issues with context. Often enough, Grammarly does not understand sarcasm, humor, or creative writing and tries to remove the first-person pronoun from texts. For example, when correcting the sentence “My brain was doing cartwheels trying to understand that lecture.” Grammarly can suggest removing “my”, changing “doing” to “performing”, or removing the saying altogether for a more direct word, such as “confused”. A sentence like “I could totally eat an entire pizza right now” can turn into “I could completely consume an entire pizza right now.” In addition to these issues, some students accept all Grammarly suggestions blindly, leading to a loss of critical thinking and identity in writing. When the tool suggests a replacement, it automatically comes with an explanation as to why the change is encouraged, yet students could simply accept without reading a word of the reasoning. Additionally, Grammarly has the “Tab” or “accept all” features, allowing users to make the changes without looking at where they are or why they are being changed. When learners accept changes without checking, the tool has the capacity to remove nuances that reflect stylistic choices. For instance, if a student decided to end every sentence with the same word to draw attention to the vocabulary choice, Grammarly would likely see that as an error and change the underlined vocabulary with synonyms, disrupting the stylistic choice. Alongside creating papers that do not entirely align with the beliefs of the writer or reflect their voice, these practices can also lead to learners never grasping the writing skills needed to write on their own. Instructors may not always be able to catch repeated errors in student writing, as Grammarly takes care of these prior to any inspection, leading to skill loss. Accepting every suggestion is clearly an issue; however, when the behavior is scaled up to the suggestions by the generative AI, it can turn into a systematic problem in the field.

Grammarly Pro and AI Writing

Similar to most chat-based generative AI tools, Grammarly AI is exceptionally fast, generating whole paragraphs and emails within minutes, which can save a lot of time, and unlike most other tools, can be easily embedded into apps such as Google and Microsoft to underline errors in every text a learner can write without checking the tool. It can also help students with writer’s block by suggesting connecting sentences as the writer requests. For instance, when being asked to generate a sentence between the benefits and pitfalls of students using their phones during class, it comes up with useful links such as “They can serve as valuable resources; however, they also have significant drawbacks,” which can work in most essays seamlessly. It is also possible to adjust the tone for more formal or informal writing. For instance, when asked for an example linking sentence to be more informal, the tool came up with the following sentence: “They can be helpful tools, but they also come with some big downsides.” These upsides, however, do bring an array of issues with them, as would be expected for most generative tools. Grammarly AI cannot always understand deep meaning, tone, or creative nuance, nor can it ideally recognize cultural or contextual errors. For instance, when asking generative Grammarly to write a five-sentence email asking for an extension, the output is very impersonal and somewhat robotic, as seen here: “I hope this message finds you well. I am writing to request an extension due to [brief explanation of the reason]. I believe an additional [number of days] would allow me to complete the work to the best of my ability. I appreciate your support in this matter. Thank you for considering my request.” These sentences can be a reminder that the tool is not always contextually aware, nor is it a creative writer. At times, explaining the context to AI can be more time-consuming than simply writing the email with one’s own writing skills. These concerns are not unique to Grammarly alone, as none of the current AI tools can think as humans do, which inevitably leads to a more robotic tone, showcasing that these tools are not producing truly original writing but rather simply predicting words based on patterns.

Pedagogical Concerns and Suggestions

In addition to impersonal texts that may carry unintended meanings, there is a larger concern at hand, which educators and researchers alike discuss: Is AI writing cheating? Some agree that it is, looking to ban the tools entirely. For instance, philosophy professor Jeremy Weissman (2023), writing in Inside Higher Ed, warns educators that AI tools such as ChatGPT represent a “novel threat to human intelligence and academic integrity,” petitioning for a return to handwritten assignments. Others find it nearly impossible to detect accurately, leaving the decision up to the learners themselves. This view carries merit as the current AI detector software has a tendency to produce false positives, with experts calling it not much better than taking a guess at times (TRAILS Institute, 2023). Furthermore, the approach may be considered mainstream since, when surveyed, sixty percent of college students reported that they were not provided with guidance on how to use AI tools in their classes responsibly (Welding, 2023). However, there is more at stake than grading, since relying on AI writing can dull writing skills and even discourage critical thinking, making education on the matter rather crucial. Therefore, a third approach has been on the rise: to educate students on ethical and effective use of AI in writing, which is utilized in my courses as well. Two or more class hours are utilized to showcase that AI use is on a spectrum from no use to asking the tool to write it for you and submitting the text as it is. For clarity, while real-life use may include more layered uses,  one approach would be to pinpoint several distinct points within the spectrum to see the progression. On one end, we would have students who complete their texts entirely on their own, which would be followed by those who utilize tools such as Grammarly to proofread their texts for grammar errors alone (i.e., only paying attention to the errors marked with red on Grammarly). The next point would be for writers who also use these tools for editing their original texts, but this time looking at issues beyond grammar, such as style, wordiness, and clarity (i.e., correcting errors underlined in all colors in Grammarly Pro). Students who dabble with idea generation would be next, perhaps asking an AI tool to outline ideas for them and keep the suggested skeleton. These learners would come right before the fifth point, which would be for students who ask these AI tools to create sentences or paragraphs while still writing on their own as well. Finally, learners who give the prompt and rubric to the AI tool and copy the response would be placed at the other end.

Through such initiatives, students are included in the conversation about their AI use and classroom policies regarding it to ensure that while they use AI as an aid, they do not allow these tools to compensate for writing skills or critical thinking. The goal is to embrace these generative tools on our own terms and limit them as needed for the outcomes of the course. For instance, for an academic writing course focused on genre education, limiting AI when it comes to grammar or vocabulary checks may not be the best idea for learners, as the course is not focusing on these skills. However, limiting these uses for a grammar accuracy class is necessary, since in such cases, grammar is the skill outcome of the lesson. As can be inferred, the rules when it comes to AI use change based on the course, the teacher, and the learners; however, as a general rule of thumb, educating learners on responsible tool use is a goal adaptable for almost all environments. For Grammarly, this training may include points like informing the learners about choosing their domains, types, and formats under the goals tab, or encouraging them to have an AI disclosure at the end of their references page. For instance, the following can be a simple model for writers to follow: I have used Grammarly Pro with the generative tool turned on/off for this paper to check for [insert reason]. I take full responsibility for my work, ensuring that it is my own. In my own classes, I do not discourage students from utilizing Grammarly or Grammarly Pro. Instead, I encourage them to take the time to engage with the tool in a meaningful manner, which looks like initiatives such as reading the reasons behind the suggested changes to see if the change is sensible in the first place, not relying on generative Grammarly to an extent where the text is missing the style and personality of the author. All in all, AI use in writing is a heavily debated issue, and educating our learners on how these tools operate, what they can and cannot do, while asking them to disclose their AI use, can be one of the most effective approaches to move forward.

Final Thoughts: Teaching in the Age of AI

Grammarly and similar AI writing tools are seemingly here to stay. As time goes on, they are getting harder to detect and more and more embedded into the writing process for writers. While it is sensible to worry about what this means for writing instruction, the best path forward is seemingly to get ahead of it. Grammarly is helpful, but it is not perfect; it can be an aid but not a tutor. Instilling the idea of responsible and effective use of these tools to learners prior to students becoming entirely reliant on them is ideal. That means teaching students not just how to use Grammarly, but also how to think with it, as an aid. How to question its suggestions. How to spot when the tool is helpful and when it’s flattening their words or rewriting them into something they were not meant to say. Critical digital literacy is now part of writing instruction, whether we signed up for it or not, and as AI evolves, education must adapt with integrity and transparency. There is a balance to find where these aids and tools are kept in the classroom while messy drafts, author voice, and passion for writing are preserved. At the end of the day, we can teach students how to write with Grammarly in the room, as long as we do not give it the last word.

Note: I have used Grammarly Pro with the generative tool turned off for this paper to check for grammar mistakes. I take full responsibility for my work, ensuring that it is my own. 

References

Lytvyn, M. (2022, November 9). A history of innovation at Grammarly. Grammarly Blog. Retrieved May 27, 2025, from https://www.grammarly.com/blog/company/grammarly-12-year-history/

TRAILS Institute. (2023, June 20). Detecting AI may be impossible. That’s a big problem for teachers. University of Maryland. https://www.trails.umd.edu/news/detecting-ai-may-be-impossible-thats-a-big-problem-for-teachers 

Weissman, J. (2023, January 17). ChatGPT is a plague upon education. Inside Higher Ed. https://www.insidehighered.com/opinion/views/2023/02/08/chatgpt-plague-upon-education-opinion 

Welding, L. (2023, March 17). Half of college students say using AI tools on schoolwork is cheating or plagiarism. BestColleges. https://www.bestcolleges.com/research/college-students-ai-tools-survey/