I don't want to get too philosophical, but I think one of the most beautiful things about humans is our ability to hold two opposing viewpoints at the same time. It's one of the things that makes human writing so difficult for ai to replicate: we are creatures of contradictions and nuance, and machines have a hard time understanding that.
All of this is a roundabout way of explaining my feelings when it comes to Grammarly’s ai writing assistant. I have been using Grammarly's spell checker for years and encourage all of my students to do the same.I was intrigued by the launch of an ai assistant in the fall of 2023 and finally decided to try it out myself.
Using the free web version that allows 100 prompts per month, I asked him to critique writing work I created, asked him to help me respond to actual writing prompts I use with my students, and asked him to respond to real student writing examples.
After doing this, I have my doubts. When working at its best, Grammarly’s ai assistant comes closest to living up to the potential of generative ai tutoring, providing supportive, thoughtful, and often very good feedback on students’ work.
On the contrary, when it misses the mark, it does so broadly, offering vague assurances and, worse, making it easier to use ai to cheat. I would argue that it could create a strong temptation to cheat even among students who turned to the tool for help without that intention initially.
Let’s take a closer look at Grammarly’s ai writing assistant.
<h2 id="grammarly-s-ai-writing-assistant-the-good”>Grammarly's ai Writing Assistant: The Good
ai%23sectionMultiColumn_6l2rsbbIfPAeGFZo8c0gWc#sectionMultiColumn_6l2rsbbIfPAeGFZo8c0gWc” target=”_blank” data-url=”https://www.grammarly.com/responsible-ai#sectionMultiColumn_6l2rsbbIfPAeGFZo8c0gWc” referrerpolicy=”no-referrer-when-downgrade” rel=”sponsored noopener” data-hl-processed=”hawklinks” data-placeholder-url=”https://shareasale.com/r.cfm?b=864277&u=1338591&m=26748&afftrack=hawk-custom-tracking&urllink=https%3A%2F%2Fwww.grammarly.com%2Fresponsible-ai%23sectionMultiColumn_6l2rsbbIfPAeGFZo8c0gWc#sectionMultiColumn_6l2rsbbIfPAeGFZo8c0gWc” data-google-interstitial=”false” data-merchant-name=”grammarly.com” data-merchant-id=”29339″ data-merchant-url=”grammarly.com” data-merchant-network=”Generic”>Grammarly bills its ai assistant as a tool that will aid in the writing process by providing assistance with tasks such as brainstorming and idea generation. And when you successfully stick to this type of usage, it is, quite simply, amazing.
I took a story I’d written for another publication that I felt was pretty close to being ready to submit and popped it into Grammarly’s web-based document editor. I then asked Grammarly’s assistant to make suggestions by clicking on the “assistant” and “generative ai” options in the menu. It gave me several good suggestions and pointed out a real weakness in the story that I was ignoring: I hadn’t provided enough specific background information about two companies mentioned.
I also submitted portions of two actual student-written papers for that assignment (both papers were partial and did not contain any information that could identify the students). I would have graded one of these papers at 90% and the other at 70%. In both cases, Grammarly provided me with encouraging and helpful feedback. In the case of the weaker paper in particular, it pointed out several of the missing elements that I or another instructor would likely notice when grading.
Grammarly's ai also initially impressed me with its honesty when it couldn't help me. For example, I asked it if it could make sure my article was written according to Associated Press style guidelines. It politely told me it couldn't.
This is great because the sign that an ai tutor is smart (or a smart person) is that they admit that they don’t know everything. Unfortunately, they weren’t always so honest about what they could and couldn’t do.
The bad
When I asked Grammarly’s ai assistant to confirm that I had spelled a person’s name correctly and consistently throughout history, the ai confidently assured me that I had, even though I hadn’t. I’m not sure if this was a true hallucination or if it simply couldn’t figure out the question. Either way, it wasn’t good.
More importantly, when I included a prompt used for one of my classes and asked for help writing an essay based on it, it provided me with an “example” by writing the essay for me. While getting an example of an essay based on a prompt could be helpful, this was more or less writing the essay for the student, which is exactly what ChatGPT and other ai tools do for students. I was hoping Grammarly’s ai assistant would prevent this.
Furthermore, it is easy to imagine how a student who asked for help and received a completed essay might be tempted to simply hand in that essay.
The ugly one
The more I analyzed these negative aspects, the worse everything became.
Grammarly's ai assistant insisted that I had spelled a source's name correctly even when I specifically asked it to check the spelling mistake I had intentionally placed in the story.
And in a disturbing and somewhat humiliating turn of events, after I put my post into Grammarly’s ai assistant, I realized that my concerns about some of its features were not hypothetical. For example, this week (I’m teaching a course this summer) I’m almost certain that at least two of my students generated entire essays using Grammarly’s ai. I also think it’s likely that many of the instances of ai-submitted papers in the past that I’ve seen in my classes were generated using Grammarly’s ai assistant. This is particularly disturbing since I had recommended all of the students use Grammarly. Did I inadvertently push them over to the dark side?
Sure, some of these issues can be overcome. In the case of school accounts, perhaps Grammarly’s ai assistant could send copies of the examples generated for students to instructors. Or better yet, it could be programmed not to generate full essay examples. ai engineers could work more closely with writing instructors to help address common needs and avoid mind-bending.
Grammarly’s other features have long impressed me, so I’m confident that Grammarly will make improvements in the future. I hope this happens soon because if we remove the problematic features from this tool, we’ll be left with something that truly delivers on the educational promise of generative ai.
