…by wrestling with both the possibilities and limitations of AI in shaping our voice in a way that respects our individual authenticity, I believe we can do as we always have, and harness the power of words.

Katie Powell

I, like many writers, am scared of AI taking my voice (Mallette 2024, Meyer, 2023). I often picture the scene in The Little Mermaid, where Ariel signs her voice away due to her own desire to become a human and, with a glowing light, watches it leave her body and sink into a jar. So while I worry about being under a kind of spell, or contract, I remain intrigued at the idea of AI harnessing my voice, of being able to tell me my voice, of potentially pretending to be me in a better way than I can be myself (Reid, 2020). That’s why I felt led to download Grammarly, one of the leading programs in AI writing assistance (though there are many that continue to surface each month), and ask it to begin tracking me.

As an English professor, I wasn’t particularly worried about the grammatical errors and suggestions it might make, but was intrigued at the way it promised to quantify and study my writing. I receive weekly reports that share the total words I’ve written, where that progress stands with other Grammarly users, and how they might categorize particular tones within my writing. Does this quantification make writing tools like Grammarly ghostwriters? Writing assistants? Critical researchers?

This piece serves as a self-guided reflection on voice, particularly the ways in which AI tools like Grammarly are able to harness my voice, what it means to have a “ relationship” (even an artificial one) with an individual long enough to learn to mimic their voice, and the possibilities that capturing something as hard to pin down as voice can have on writers. With Grammarly as a case study, I work to reflect on the idea of AI as a writing tool, diving into the oft-muttered suggestion (and fear) that AI can become a kind of replacement for a writer and their voice. Instead of a replacement, I am using AI to learn about my voice. I believe (or really, I choose to hope) that leveraging AI to help us better understand ourselves and reflect on our own voice shifts such controversial technologies from objects of debate to a means of critical reflection.

By engaging with this tool and reflecting on each aspect of the deliverables (going into the belly of the beast, if you will), I plan to emerge with a case study and strategy that others can use to more accurately study their own voice and to better understand the power and potential of AI not as a replacement, but as a mirror of our writerly selves we might not be able to see otherwise.  

AI as Storyteller

I wonder if we can use these tools to individualize our own voice and see AI as a mirror to harness (and therefore tell) our own unique stories.

Katie Powell

I downloaded Grammarly in February 2024, only because it was a requirement of a freelance gig. Since then, Grammarly claims it has analyzed 22,404,066 of my words. It claims I am more productive than 98% of their other users. I would imagine, based on the students I know who use Grammarly, that these sorts of tools aren’t often used for people who especially like to write, who want to be writing, who hope to continue to make a living based on being good writers. The clear emphasis in these initial reports is that of a productivity narrative, a prioritization on being grammatically correct and professional. This particular direction toward productivity reinforces a shift away from love of writing, and instead a shift into what the world interprets as “good writing,” and of course lots of writing, even if the words penned are not of stellar quality.

Even now, as I work on this article, I have a suggestions tab open to the righthand side, displaying each and every word I write (that doesn’t interact with student information) that is constantly shifting as it offers suggestions, clarifications, and ideas for clarity, for tone, for confidence. In this way, am I like Ariel, signing away my voice in a sketchy underwater contract?

While these numbers feel staggering, I was eager to consider the potential data that could be drawn from such a close and comprehensive view of my writing. Though I have always struggled to articulate a particular ethical perspective around issues of AI, this level of intimacy in my work and in my writing leaves me unsure of the normative ethics, establishing the right and the wrong, of what to do with this information. And just as I’ve asked my students to explore and expose themselves to AI to begin to establish a code of ethics, I hope to use this piece to do the same. It is clear that there is real student and professional panic around the opportunities that AI presents and what it might mean for our jobs.

My “day job” is professional writing professor, so I am first and foremost concerned with student reactions to this idea of a loss of voice. I want to ask our students to “lean in” to these opportunities with a clear and earnest consideration of social justice, accessibility, and advocacy issues (Mallette, 2024).  Additionally, I center and lead with the notion that understanding both AI’s capabilities and limitations will allow students to better harness both the potential of AI and their capabilities and understanding within it.

I, like so many of us, remain concerned that in the process of encouraging students (and myself) to lean into the possibilities of AI, we are losing a critical sense of voice. One of my greatest joys, and I believe many of our collective greatest joys as writing teachers is not just getting to know students in class, but learning to hear and recognize their voice throughout our time together. While this recognition can often be somewhat intangible (Nelson and Castello, 2012), or focused on relationships (Elbow, 2007), it remains a critical part of our work as writing teachers.

So what becomes of us when we allow ‘the voice of the other’ to write for our students (Meyer, 2024)? By this point, we’ve all seen student papers, news articles, or advertisements that are clearly AI generated. These often more lifeless/soulless pieces feel generic and underdeveloped. And yet they seem to speak again to this idea of “perfect” or productive voice that Grammarly is a product of, and has even begun to recreate for its users. Whose voice are we really talking about when we talk about seeking an authentic voice in professional and academic contexts?

Many scholars, particularly scholars of color, have pointed to the inherent whiteness, or dominant culture focus, on the outputs of AI. As Carmen Kynard (2023) puts it in her own blog, “School literacy, white English, academic discourse, and bourgeois western rhetoric/education are so standardized and robotic that robots do it better now. Robots have come home to their roosts and I don’t feel sad or scared about none of it.” These conversations mirror discussions we’ve had of the cultural nuance in defining professionalism (Gelms and Johnson 2024), and of course the question of who is a professional academic (Hull, Shelton, and McKoy 2020). 

If AI is a reflection of us (which we’ll review as I consider “AI as mirror,” below), it becomes clear that the soulless pieces we’ve all detected from AI are merely a product of the standard “professional” pieces we have taught it to value. And so we can’t talk about the idea and the hope of students cultivating their own voice without talking about whose voice is inherently prioritized.

 Yet I find myself curious about the possibilities of AI, and feel a bit of a tension between asking students to explore it, and finding ways to combat it. For professional writing professors like me, part of my job is not just this larger focus on helping students find their voice, but helping them to channel their own unique “professional” voice, whatever that means. This focus on the “professional,” I worry, lends itself to AI abuse even more. Students (and the public) seem under the impression that the professional voice of technical documents, marketing materials, and think pieces is even more likely to be generated, produced, and recreated by AI. Is giving AI the power to produce documents akin to Ursula, dressed up as Eric’s enchantress, mesmerizing all who will slowly forget the human (or, in this case mermaid) behind the story?

Amidst this crisis of loss of voice, there is a real push to tell stories, an uptick in the need for story, or (my interpretation) a human connection behind the clear and careful writing that comes in a professional space. A Wall Street Journal article on storytelling claims that “The percentage of LinkedIn job postings in the U.S. that include the term “storyteller” doubled in the year, to include some 50,000 listings under marketing and more than 20,000 job listings under media and communications that mentioned the term, according to the professional-networking platform.” (Deighton, 2025). Storytelling, though a buzzword, further implies (rightfully so) a personal connection, a human approach, a uniqueness that AI will never be able to replicate. In this way, I believe that story can act as a counter to the generic professional voice prioritized in professional communication of which the standard AI voice is a direct result. It is clear by the commentary from thought leaders on LinkedIn and the writing in these job ads that there continues to be a place for real, human connection amidst the large and daunting promises of AI.

As a professional writing professor, it can be hard to consider myself a storyteller. The ability and the power to craft words and narratives is powerful and upending, something I believe will continue to buoy me into the future, as it buoys my students. Despite my temptation to focus simply on my students, my pedagogy, and my job to cast either doubt or support for AI, part of being a storyteller means telling my own story of AI use, my own explorations in allowing AI to consider my voice. So I downloaded Grammarly. For better or worse, I’ve allowed it to watch me for the past year, to study my habits and listen in on the ways in which I write in different genres and mediums. And I hope and want to use it to be a more thorough and informed writer and tell a more powerful story. Therefore, I’ll share with you the ways in which it has both recorded me and begun to attempt to channel my own writing style and tone.

AI as Mirror

This kind of mirror approach allows me to question the tones I hope to bring to any writing project, and perhaps drill into my work in Grammarly to establish those tones.

Katie Powell

I believe we can use the power of AI and the renewed focus on storytelling to become better storytellers and writers, to act as a kind of mirror to our own writing. I suppose I argue that AI is a kind of mirror of our prioritization on the dominant culture in professionalism. However, I wonder if we can use these tools to individualize our own voice and see AI as a mirror to harness (and therefore tell) our own unique stories. Does the “relationship” tools like Grammarly establish with us combat the general bias that AI writing produces? Though I’m interested in how this will help my students, I decided to start by “practicing what I preach” and turned a quest for voice, authenticity, and story on to myself.

I receive an email report about once a week measuring my progress on Grammarly. This report shares the total number of words I have written, and (importantly) compares those words to other users on Grammarly. It wasn’t until reflecting for this article that I realized the significant emphasis Grammarly places on tone. Based on these weekly reports and the headlines they choose to use, tone and progress seem to be the key indicators of writing, or of what they offer to talk to us about writing here.

For instance,

• My report on 12/2 shares that I wrote in a unique tone.
• On 12/8, I was encouraged to “take a look at your most used tones below.”
• On 12/15, I was applauded for making progress,
• On both 12/29 and 1/5, I was complimented for using “fresh tones” as I worked.

These headings are complemented by the numbers: my “writing streak,” how to get to an achievement badge, and information on my productivity (how many words I typed this week), my accuracy (how many alerts I received on my grammar), and my vocabulary (how many unique words I used in my writing). In each of these, I am given a comparison percentage to other Grammarly users to determine “where I stand.” I am then given a list (with emojis) of the tones that my writing takes on and how that stands compared to previous weeks and writing:

I initially found “defensive” and “disheartening” to be amusing tones to detect. However, those tones have continued to occur and come up as listed as I move through my time with Grammarly.

One significant note is that it is unclear, at least to the user, is how these tones are being detected. I am not given, for instance, a sense of the “coding” that is perhaps being used to determine each of these tones.

In future analyses, I hope to engage with developers at Grammarly to uncover if these tone determinants are being made at the level of individual words, sentences, or paragraphs, though I would imagine much of that information is proprietary. As we might infer from these listed tones, much of the identification is subjective based on culture or rhetorical tradition. “Defensive,” for instance, feels like a particularly gendered term.

But does Grammarly know my gender? Or my nationality? While I can’t answer these questions, I can explore and consider my story behind these words and reports. By investigating what was going on emotionally that propelled my shifts in tone, my various use cases, and how those played out in my writing, I can learn more about my own voice in different instances and situations. For instance, the below chart of tone is from September 8th.

My confidence level is higher, I’m more direct, friendly isn’t even listed at all, and I’m only 10% appreciative as compared to the 30% appreciative I received in December. When I look back on the writing projects I was working on at this point in the semester—beginning to craft a proposal for a research grant on AI and ethics and professional writing, writing a first draft of an article for a special issue focused on “hope amidst crisis” and preparing my first few weeks of communication for a first year seminar meant to acclimate students to the University of Cincinnati campus—these tones make sense. Confidence is necessary in a research proposal, so this report confirms the proposal draft exudes that confidence. I use the literal word “hope” 24 times in my special issue article and the word “believe” 8 separate times- perhaps that translates to confidence? Additionally, I’m glad to see that they detected a directness to my writing- I have been working on streamlining my syllabus and course policies in order to make the language clear and accessible to students, particularly first year students.

Despite the lack of evidence revealing how Grammarly came to the conclusions it made on my tone, I do believe that these varying tones indicate for me the heavy presence of emotion at work in my writing. Writing at any stage is an emotional process, and I see Grammarly’s reports on tone offering up a reflection of that. In this way, perhaps, Grammarly is used less as an author, and more as an observer, a quantifier, a tool that provides a bit of insight and depth into my emotions, my patterns, and the larger message I am attempting to send.

Another potentially emotional component of this, however, is the relationship (if we can call it that) that Grammarly has developed with me as it learns my patterns, my voice, my tone. This long-term observation, Grammarly claims, allows it to understand a bit of me. Based on the fact that this observation, or this relationship, is artificial, is this sort of relationship ethical? And if it can “objectively” observe me (though it’s difficult to call the tones it quantifies objective), how does this shape the way I view and understand my “authentic” voice, the very idea of authenticity?  

As is clear through my shifts in tone based on my emotions throughout the semester, my voice is cultivated through intentional practice. AI-supported strategies, such as these Grammarly reports, can therefore help with the goal of becoming more aware and intentional about my authentic voice. By holding up my Grammarly reports as a kind of mirror on my own writing, I do believe I am better empowered and informed on making changes and shifting strategies depending on the writing project and the time in the semester.

This kind of mirror approach allows me to question the tones I hope to bring to any writing project, and perhaps drill into my work in Grammarly to establish those tones. When I work on course materials, freelance writing, or traditional technical communication scholarship, for instance, I seek the confident, informative, and formal tones listed in the report. But in my more personal writing, perhaps even as I wrestle with incorporating story throughout my scholarship (including in this piece), I seek a curious, friendly, or what Grammarly would define as an informal tone.

Offering me a summary of how productive and accurate I am, as well as what level of vocabulary I offer, provides a kind of motivation that I think works for me. While it prioritizes the amount of words and the individual words used, as well as grammar and vocabulary over quality of writing, having these metrics and then using actual humans to quantify and assess my own writing is more fitting anyway. I could certainly see this numbers-based motivation working (in part) for our graduate students or even productive and motivated undergraduates. If I’m being honest, it’s weirdly helpful to know that someone is keeping an eye on and assessing my writing, and useful to see the ways in which they are able to reflect what I am going through emotionally both in the semester and as a writer. The role of Grammarly in this case, then, is an informed and data driven mirror, offering me an opportunity for reflection that I personally do indeed find helpful. But an important question I must ask myself is: does using Grammarly as a data driven mirror reinforce the dominant culture implications that can often be part of the idea of voice? I decided that I needed to allow it to write in my voice in order to begin to answer this question.

AI As Writer

Thus far, I believe that reflecting on my Grammarly reports has allowed me to see Grammarly and its possibilities as more of a mirror, instead of a direct author. However, I also wanted to ask Grammarly to produce something for me. I opted to use this very piece, thinking I might like to reflect more broadly on tone and the priority that Grammarly places on it within my weekly reports. Therefore, I’ll address a few different iterations of my prompting within Grammarly. Within the blocks of text it produced from my prompting, I found that one key illustration was the most illuminating of the power of this kind of technology and how it might help us question authenticity and consider how we might use these sorts of technologies.

I first opened my site navigation and asked it to write “2 paragraphs on the importance of tone in writing.” After some of the usual bland yet thorough introductions, it offered up an example by saying

I then said, “can you make it sound more like my style based on what you have observed in my writing?”

It responded with:

In regards to the sentence focused on conversational tone, it thinks I am more likely to use approachable language like ‘break down barriers’ instead of “make complex topics more accessible.” I like its change from inviting readers to “explore ideas” to inviting readers to “engage with our thoughts.” I have to hand it to the software- that is much more me. Could this author be the dark-haired Vanessa, passing as the woman of my dreams while remaining Ursula at heart?

From there, I followed Grammarly’s prompting to select an option to “make it empathetic,” which it edited to:

And then finally, I clicked on “make it inspirational,” and it produced the following:

Both of these results came back, not surprisingly, very similarly. I really like the shift here at the beginning. Instead of saying “adopting a conversational tone” it offers a much more active and human-centered “when we adopt a conversational tone” that is, indeed, far more empathetic and inspirational. In fact, I’m not certain I would think to offer up this kind of example to my students, but I thoroughly intend to in class. However, having direct comparisons to the same basic language provides a more bird’s-eye view of our own writing. Additionally, asking Grammarly to generate this text in my style allows me to see the ways in which I typically write, and the conscious and subconscious decisions I choose to make. Pairing with my beginning-of-semester/end-of-semester tone list, “inspirational” and “empathetic” didn’t even make the cut. I can use these instances of what empathetic or inspirational looks like in action, should I seek to add that tone back into my work. Additionally, I can think critically about where this tone sits for me individually, or how it combats a “standard response.” Does the inclusion of “we” or the subtle shifts into empathetic or inspirational actually work for storytelling? And when we see these shifts as positive, are we ultimately combatting the standard language for more individual personality?  I like to hope so, and I do believe Grammarly’s suggestions can work to make me a better writer.  

Conversely, of course, drilling down to Grammarly’s interpretation of “empathetic” vs “accessible” offers us a look at the mostly surface-level use of these tools. The only main difference in these tones is the use of “approachable” for empathetic and “accessible” for inspirational. What an interesting discussion could be had with students, or with myself, to think about the ways in which Grammarly has been studying me and my writing, coming to the conclusion that I (or we as a society) think of accessible as more of an inspirational (perhaps even aspirational) stance, while approachable allows us to demonstrate empathy. When asked to write this sentence in my style (see above), it chose to use approachable instead of, as in other instances, accessible. This potentially signals a lack of depth, simply using a formula that interchanges “accessible” and “approachable” at random. Even if it does, however, I think the questions it brings up, and the reflections it offers for me, are worth delving into.

AI As Support

I’ll conclude this thought piece by running through a sentence from my first draft, showing each iteration Grammarly produced to adjust the tone:

At the end of The Little Mermaid, Ursula’s spell on Ariel and on many others is broken first when the necklace that contains Ariel’s voice is broken, and then again when Prince Eric rams his ship into her side. In 2026, however, I’m not certain there is a ship big enough to ram and stop the rapid development of AI tools and technologies. However, I do believe there are ways to use these voice-capturing tools to our advantage, not to our demise. These concluding sentences represent a world of possibilities for our students, or for ourselves.

What power does having these variations have on our decision-making? What does Grammarly’s own choice of tone say about what we have taught it about tone? Instead of positioning Grammarly as something to either embrace uncritically or reject outright, I hope to use it to support metacognition and self-awareness. I believe that having a larger, chartable view of the tones it thinks I take gives me more insight and more depth, even while the actual detection is likely surface level. This, to me, gets at the very heart of our fears about AI: Is it taking our voice, or can we use it to better understand and refine our authentic voice? Does an individual relationship with an AI tool that “learns” your style combat the standardization of voice AI supports, or does it encourage it?

While AI offers a mirror to reflect on our writing, it also raises ethical questions about authenticity and control. By exploring this tension, by wrestling with both the possibilities and limitations of AI in shaping our voice in a way that respects our individual authenticity, I believe we can do as we always have, and harness the power of words.

References

Deighton, Katie. “Companies are Desperately Seeking ‘Storytellers’.” Wall Street Journal, December 12, 2025. https://www.wsj.com/articles/companies-are-desperately-seeking-storytellers-7b79f54e

Elbow, P. (2007). Voice in Writing: Embracing Contraries. College English 70 (2), 168-188.

Gelms, B., & Johnson, C. (2024). (Re)defining “professional” in technical & professional communication. Technical Communication Quarterly, 33(3), 249–257. https://doi-org.uc.idm.oclc.org/10.1080/10572252.2024.2342581.

Hull, B., Shelton, C. D., & Mckoy, T. (2020). Dressed but not tryin’to impress: Black women deconstructing “professional” dress. Journal of Multimodal Rhetorics, 3(2), 7-20.

Kynard, Carmen. (2023). When Robots Come Home to Roost: The Differing Fates of Black Language, Hyper-Standardization, and White Robotic School Writing (Yes, ChatGPT and His AI Cousins). Education, Liberation & Black Radical Traditions for the 21st Century, http://carmenkynard.org/when-robots-come-home-to-roost/.

Mallette, Jennifer. 2024. Preparing future technical editors for an artificial-intelligence-enabled workplace. Journal of Business and Technical Communication, 38, (3), March 2024, 289-302.

Meyer, Craig A. 2023. Generative AI: the voice of the other. Double Helix, Vol. 11 (2023).

Nelson, N., and M. Castello (2012). Academic writing and authorial voice. In M. Castello and C. Donahue (Eds). University Writing: Selves and Texts in Academic Societies (pp. 33-51). Emerald.

Reid, Alex. 2020. Synthetic speech and negotiation: AI’s nonhuman rhetoric. Enculturation (June 2020), https://enculturation.net/synthetic_speech_and_negotiation.