Recently, I read how some are assuming that metaphors and similes are proof that AI wrote the manuscript. My eyes rolled back into my head so far that I saw a 10-year-old me with a “WTF” look on his face. That is perhaps the most asinine AI argument I’ve heard to date.
After I carefully rolled my eyes back forward again, I decided to do some digging. Are these claims factual, or are they simply ragebait posts on Threads that someone posted to fuel engagement?
Because…social media.
Who Is Saying to Remove Metaphors and Similes?
After digging into the claims through social media posts, blog content, and even Reddit, I can’t find a single instance of an actual editor, agent, or publisher demanding that authors remove metaphors and similes due to the threat of being labeled as AI writing.
Instead, what I found was that editors and other authors are mentioning how basic metaphors and similes should be avoided. But not because people will think it’s AI, but more of improving writing skills. Keep in mind that you want to show, not tell, a story.
However, AI will often use things like “smells like,” “looks like,” “acts like,” and more, as it is a predictable sentence structure. Usually, the AI will also include a lot of nonsensical comparisons, as it lacks the capacity to deliver actual insight into a scene.
In other words, it lacks emotional connections to pull ideas from the reader’s mind.
So, yes, editors and others are telling authors to remove cliches and weak writing elements in metaphors to avoid AI claims while delivering a more engaging tale. It’s not like they are telling authors to remove them altogether.
Now, there are times when using smells like or looks like is an apt description within your story. What professionals are saying is to limit the use of those kinds of similes. Be more descriptive to engage the reader’s imagination.
This is a perfect example of being triggered by an anecdote. But, as I said before, I am open to seeing evidence to the contrary. At least, evidence that goes beyond the anecdotal.
Don’t give me the “But so-and-so said so, it must be true” argument.
What Are Metaphors?
A metaphor is a figure of speech that is a direct comparison of two unrelated things to bring about a specific meaning or understanding. The metaphor is more of an absolute comparison by insinuating the two things ARE the same.
An example of a metaphor would be: “Love is a cannibal.”
This means that love can be an all-consuming affection, devouring your soul and very essence of who you are. It can also be an unhealthy obsession, as well as a drive for intense intimacy.
It’s also a song by Sir Elton John on the Ghostbusters II soundtrack.
Creative writing uses metaphors quite often to get the reader’s imagination involved. This is especially true in poetry, as a lot of writing focuses mostly around metaphorical phrasing, rhymes, and/or depictions.
You probably use metaphors on a regular basis and don’t even realize it. How often do you claim that “coffee keeps me from getting stabby?” Can coffee really keep someone from stabbing another? Not really. It means that my mind isn’t fully socialized until after it’s been caffeinated.
In other words, you’re feeling a bit more aggressive before you had a chance to wake up, which happens a lot at my age.
What Are Similes?
A simile is a figure of speech that directly compares two things, focusing on how they are similar, regardless of absurdity. The idea is to stir an image, smell, or feeling in the reader to accentuate the moment.
An example of a simile would be: “The children fight like cats and dogs.”
This means the fighting is often, and sometimes brutally, as cats and dogs are historically enemies of one another. This is a bit of a misnomer, though, as cats and dogs often get along very well. But it’s the visual of not getting along that is why it is used.
Another example would be: “She smelled like dead fish and seaweed.”
A disgusting thought, but a simile nonetheless.
Like metaphors, similes are often used in creative writing, whether you’re writing prose or poetry. They stir ideas for the reader to get a sense of similarity within the lines.
Let’s take the example above; You can easily envision cats and dogs fighting, so it’s not a far leap to say that you can envision the children fighting as well.
Like, as, and than are often words associated with presenting a simile. It just depends on how you put it together.
AI Trained from Original Work!
When you see claims on social media about “surefire” ways to determine AI writing, keep one very important thing in mind. Artificial intelligence was trained on the works that already exist. If you see AI using an em dash, that’s because authors use an em dash.
Of course, nowadays, you also have some AI platforms learning from other AI platforms, which is kind of ridiculous, really. It reminds me of that scene from iRobot where Will Smith’s character, Detective Spooner, sees the manufacturing plant.
“Robots building robots…well, that’s just stupid.”
Anyway, to say that one can detect AI because of punctuation or grammar usage is north of asinine. This is especially true when you consider that AI is constantly improving…sadly.
It doesn’t help that approximately 54% of adults in the United States read below a sixth-grade level. This helps push the narrative that certain things are AI-created. If someone doesn’t know what an em dash is, they’ll automatically assume it was generated by AI, for example.
Now, there are some signs that AI may have written text, such as poor word choices, nonsensical metaphors, or repetitious verbage. However, there are also a lot of newer authors who implement those elements because they don’t know any better. So, even those things are not an absolute guarantee.
For instance, I often have to go through and remove a lot of repeated phrases and redundancies after the first draft. It’s just not something I think about while writing, as I am just trying to get the story out of my head.
The point here is that generative AI is trained on the works of real authors who came before it. There is a reason why the AI decided to insert certain metaphors, similes, em dashes, and whatever else people think is bot-driven.
Is AI Detection Faulty?
While I was experimenting with plagiarism detectors, I tested out a few of the AI detection tools available. What I found was that many of them flagged my originally written piece as AI-generated. You can imagine how incensed I felt.
Too many of them were flagging my essay, and it made me think how often teachers and professors use these tools to determine grades for students. A lot of these “AI detection” tools delivered false positives, which could influence the instructor’s judgment.
I don’t know about you, but I’d be pretty pissed if I worked my ass off on an essay only for the professor to give me a failing grade because an AI detector falsely flagged it.
Then, consider right now, AI is the worst it will ever be…meaning it’s just going to get better from this moment on. Eventually, AI detection tools will flag every piece of content because they don’t know any better. Or, every piece of content will be considered human-written. Either way, it defeats the purpose of having the tool in the first place.
When it comes to metaphors and similes, AI will eventually be able to assemble them in a way that is unidentifiable compared to real authors. It’s just a matter of time.
At this moment, though, the only way to truly tell that authors are using AI in their writing is by standing over their shoulders and watching them type prompts into ChatGPT or any other system. Because trying to judge writing by grammatical structure is not it.
Well, unless the person left the AI prompts in the book when publishing, which has happened recently.
Stop the Damn Witch-Hunts!
It’s sad to see legitimate authors get one-star reviews on Amazon because the reader suspected the book was written by AI. False accusations can ultimately kill what could have been a great career.
In addition to negative reviews, it doesn’t help when people spread misinformation on social media for ragebait clicks. It propagates a false equivalency that gets folks wound up in a tizzy.
It’s also upsetting to see legitimate essays get flagged with a potentially poor grade because of detection tools.
Does AI write poorly crafted metaphors and similes? Absolutely. But so do a lot of newer authors who can’t afford editors.
Before you submit that poor review or leave a ragebait comment on social media, think of the impact it might have. Authors, especially those of us who are self-published, already have an insane hill to climb for success. Don’t be the bottom-feeder who is constantly tossing boulders on the path.
- No, Metaphors and Similes ARE NOT Proof of AI Writing - April 6, 2026
- Why I Want to Minor in Anthropology as a Writer - April 3, 2026
- My April Writing Challenge in 2026: Blogs and Books Galore - April 1, 2026