1091. Is the em dash a sign of AI writing? I looked at where the idea comes from, and we have the final answer! Then, we look at the difference between "caregiver" and "caretaker."
1091. Is the em dash a sign of AI writing? I looked at where the idea comes from, and we have the final answer! Then, we look at the difference between "caregiver" and "caretaker."
The "'caregiver" and "caretaker'" segment was written by Jim Norrena, MFA, who has been writing and editing for more than thirty-five years. He’s the founder of TypoSuction.com, an independent editing/writing service. He's taught grammar and copyediting intensives and professional proofreading workshops at Media Alliance and served as events coordinator for Bay Area Editors’ Forum (BAEF). Visit Linkedin.com/in/jimnorrena/ for his complete work history and highlighted projects.
🔗 Share your familect recording in a WhatsApp chat.
🔗 Watch my LinkedIn Learning writing courses.
🔗 Subscribe to the newsletter.
🔗 Take our advertising survey.
🔗 Get the edited transcript.
🔗 Get Grammar Girl books.
🔗 Join Grammarpalooza. Get ad-free and bonus episodes at Apple Podcasts or Subtext. Learn more about the difference.
| HOST: Mignon Fogarty
| VOICEMAIL: 833-214-GIRL (833-214-4475).
| Grammar Girl is part of the Quick and Dirty Tips podcast network.
| Theme music by Catherine Rannus.
| Grammar Girl Social Media: YouTube. TikTok. Facebook.Threads. Instagram. LinkedIn. Mastodon. Bluesky.
Grammar Girl here. I’m Mignon Fogarty, your friendly guide to the English language. Today, we're going to talk about the AI em dash and the difference between a caregiver and a caretaker.
by Mignon Fogarty
I have been hoping this idea I keep seeing on social media that em dashes are a sign of AI writing will burn itself out, but it … Just. Keeps. Going. And it's driving me crazy. I am so annoyed that I had to drop everything and crash this piece into the editorial schedule because em dashes are definitely not a sign of AI writing, and I'm seeing too many people thinking they are. I've been using em dashes more in my writing over the last few weeks. Every time I see another social media post about it, I use another em dash!
Nobody is sure where it came from, and I can find social media posts from last year talking about it, but the conversation seemed to really surge in March, when the LuxeGen Podcast — a GenZ podcast about fashion, beauty, life advice, and pop culture — referred to it as "a longer hyphen" in a video on March 22 that got about 2.5 million total views on Instagram and TikTok. Here's the relevant part of the audio:
"ChatGPT hyphen is getting a lot of stick at the moment, OK? Pretty Little Thing recently had a rebrand, and the top most liked comment was someone being like, "I can't believe they let the chatGPT hyphen in." It's a longer hyphen. I don't know if you've noticed it." [fade out]
You can tell they're talking about something they saw other people talking about, so they didn't create the idea, but they're very likely the ones who put rocket fuel behind it. The day after this video, March 23, you can see searches for the phrase "em dash" spike up on Google Trends in most English-speaking countries, and it has stayed higher ever since.
And can I just say that these people don't even know the name for the punctuation mark? It's not a long hyphen; it's a dash. So you really shouldn't trust them when it comes to determining whether this is an ironclad sign of AI writing. You absolutely shouldn't trust me for makeup or pop culture advice, and you shouldn't trust them for punctuation advice.
And whenever I see social media posts about this, which is multiple times a day, the comments are filled with writers who disagree, and I actually think this is the reason it won't die — the algorithms like controversy. They like comments. So if a post gets a lot of comments, it gets more reach, and these posts always do.
And then people who want to get more exposure on social media see that these posts are doing well and jump on the bandwagon, which is why I've tried to avoid talking about it. I didn't want to feed the fire, but I just couldn't ignore it anymore.
So why might AI use a lot of em dashes? And if it does, then how could seeing a lot of em dashes in someone's writing not be a sure sign they've used AI? Well, all the writing you see coming out of tools like ChatGPT are the way they are because they were trained on human writing. The only way em dashes would be in there is if people used them.
Emily Dickinson used a ton of em dashes, for example. And it's easy to find posts online about people overusing dashes before the existence of PT. There's a post on a Reddit fan fiction subreddit from 2021 that asks, "Am I the only one addicted to em dashes?" with lots and lots of comments from people chiming in about how much they love and use them.
Em dashes add drama, and they make whatever comes after them or between them stand out. They can create a certain conversational tone. Lots of people use them, and some people just use them more than others. I love them.
So if the em dash isn't a sign of AI writing, then what is?
Well, there are no studies I've been able to find about em dashes being over-represented in AI writing, but studies have found that certain words do seem to show up more in AI writing. Last year, Scientific American published a piece describing a "suspicious increase" in certain lofty words in scientific papers in 2022 — the year ChatGPT came out — including "innovative," "notable," "noteworthy," "pivotal," and "intricate."
Around the same time, another group found that the appearance of the word "delve" skyrocketed in titles and abstracts in a database called OpenAlex for scholarly articles in 2022, and another analysis found it skyrocketed in papers on PubMed, which indexes biomedical papers, in 2023.
So yes, it seems likely that certain words were over-represented in AI output, at least for a while, but that still doesn't mean you can look at writing and immediately think it was written by AI if it has a "delve" or an "innovative" in it because people actually use those words too. Again, the only reason ChatGPT includes them is because people use them and they were in the training data.
Listen to my interview with AI expert Christopher Penn
One of the studies that found the tell-tale words looked at 140 million papers. And when you do something on that scale, you can see a global change. Some words seem to clearly be showing up some percent more often after ChatGPT was launched, but that doesn't tell you anything about one single paper you're looking at.
Finding these words could be the tiniest of clues, but the writer may just like the word "delve" or really think someone is meticulous or an action is commendable. We shouldn't have to give up perfectly good words just because AI seems to overuse them, and I would never accuse someone of using AI based on the presence of a few words.
And unfortunately, I also want to caution you against using AI detectors. These are tools that claim to be able to flag AI writing, but study after study has found they have unacceptable error rates, especially in certain populations.
The companies claim they have 99% accuracy rates, but a study out of Stanford found that seven widely used AI detectors incorrectly classified essays from non-native English speakers as AI 61% of the time.
Researchers have also suggested that these tools wrongly flag writing from people who are neurodiverse, and a report by Common Sense Media found that Black teenagers are twice as likely as other teenagers to be falsely accused of using AI.
I consider these tools worthless at best and harmful at worst. I'm sympathetic to what teachers are up against with AI cheating, but these tools aren't a reliable or fair way to solve the problem.
So are there any ways to identify AI writing? Well, an obvious sign is when people accidentally leave in the introductions to the text the AI has created for them, like "Here's your essay about the Civil War," which seems to happen surprisingly often (or it could be that I hear about them often because it's so ridiculous it makes for good social media posts).
One thing I've seen that seems like it could work in some cases is teachers putting hidden text in their assignments that a chatbot would see if a student copies and pastes the assignment. For example, one teacher posted about including hidden text in assignments that would cause ChatGPT to cite his dog, Frankie Hawkes.
And I think an eighth grader one day and like David Foster Wallace the next, it's reasonable to suspect that something is up.
But even dramatic changes in writing style aren't always a sign of AI writing. I recently took a class offered by LinkedIn on how to write better LinkedIn posts because I teach so many LinkedIn Learning courses, and afterward I dramatically changed how I wrote my posts. If you just saw that my posts were suddenly different, you might think I had started using AI, but that wasn't what happened. So maybe the person you suspect just took a class, started working with a tutor, or is just trying something new.
Finally, chatbots don't all write the same anymore either, and their writing has changed over time. I've heard a ton of people say they particularly like how Claude writes or that ChatGPT 4.5 is an especially good writer, so there is no one "AI style."
There isn't even one ChatGPT anymore. When I look at it right now, I can choose from seven different models. Plus, it's trivial to tell AI to write in different styles. As a joke, Daniel Heuman's team at Draftsmith made a button for Talk like a Pirate Day that would turn all the AI text into pirate speak.
So to sum up, this idea that em dashes are a sign of AI writing seems to have been thrust into the mainstream by a popular podcast and ironically, perpetuated by writers expressing their indignation with the idea on social media. There's no proof that it's true, and while there are some other ways to possibly detect AI writing, none of them are good enough to go around accusing people of it.
So please, please, please don't believe all the social media posts you see saying that em dashes are a sign of AI writing, and please don't stop using them.
by Jim Norrena
Have you ever stumbled when saying “caregiver” or “caretaker,” wondering what’s the difference? After all, at first glance, one is “taking,” while the other is “giving,” yet they’re often used interchangeably.
That’s because despite their contradictory appearance, they are in fact synonyms — words that mean the same thing.
Does the English language truly need two words to describe the same job?
Well, you might be surprised to learn a caretaker can mean something quite different from a caregiver. Now you’re probably thinking, “Wait! You just said they’re synonyms. How can they be different?”
Let’s begin with "caregiver." This is someone who provides direct care to other people, usually children, the elderly, or the chronically ill. Tasks may include bathing assistance, cooking and cleaning, driving someone to medical appointments, and so on.
Caregivers often act out of personal interest (and some out of professional duty). A caregiver can be a friend, a neighbor, or a paid helper. They typically work out of a person’s private home, in a convalescent home or hospital, or as part of a hospice team.
"Caretaker,” on the other hand, has two meanings:
First, a caretaker is “someone who gives practical physical or emotional care and support to another person.” (By this definition, a caretaker is akin to a caregiver.)
But a caretaker can also be someone "who takes care of the house or land of an owner who may be absent.” This kind of caretaker is usually in a managerial position such as warden, custodian, keeper, or guardian. Typically, the job is to manage property — trimming grass, picking up mail, feeding pets, and doing general household maintenance tasks.
Defined this way, the caretaker’s role is restricted to just things, not people. Only caretakers manage things.
To help illustrate the difference, let’s look at two examples:
Squiggly gives Aardvark his medicine whenever Aardvark is sick in bed.
Here, Squiggly is Aardvark’s caregiver or caretaker because Squiggly is taking care of Aardvark by giving him his medication.
When you’re describing someone who cares for people, you have nothing to worry about — either “caregiver” or “caretaker” fills the bill!
In contrast, the role is more tightly defined when only things are being cared for:
Aardvark is going to house-sit for Squiggly, so Squiggly can go on a much-deserved vacation.
Taking care of a person’s home — not those within the home — makes Aardvark a caretaker, but not a caregiver. Aardvark is only managing property, not Squiggly himself.
While both are examples of caretaking, only the first example illustrates caregiving.
“Caretaker” is the older term and also the broader term because it encompasses taking care of people, property, and things. Think of it as the catchall term. According to Etymonline, it first appeared in 1769.
About a century after “caretaker” appeared in the U.K., "caregiver" showed up in the United States in a book about mental health, and according to a Google Ngram search, which shows how often words appear in books indexed by Google, "caregiver" continues to be more common in the US than in the UK, which seems to have instead adopted "carer" as an alternative word to describe taking care of people instead of things.
It doesn't seem that anyone documented why they felt the need for an alternative word like "caregiver" — at least not that we can find — but some people speculate that the word is meant to be more personal, to show that caring for a person is different or more meaningful than caring for property, or to emphasize the positive associations we have with the idea of giving rather than taking.
Just remember:
With caregiving, in order to give care, there must be someone to receive care. And when you’re caretaking, you’re taking care of everyTHING.
And that pretty much … takes care of this matter!
That segment was written by Jim Norrena, MFA, who has been writing and editing for more than thirty-five years. He’s the founder and principal editor at TypoSuction.com, an independent editing/writing service. He taught grammar and copyediting intensives and professional proofreading workshops at Media Alliance and served as events coordinator for Bay Area Editors’ Forum (BAEF). You can find him at LinkedIn.
Additional Resources
Online Etymology Dictionary, s.v. “caregiver,” accessed March 5, 2025, https://www.etymonline.com/word/caregiver.
Finally, I have a familect from Don.
Hey, Grammar Girl, my name is Don. I'm in Boston heading west on US 20, which is, by the way, the longest road in America, and this is where it begins. Anyway, my daughter, when she was young, used to sometimes come in the kitchen at breakfast time and say that she was a medium bird. Okay, what's a medium bird?
She said, "Well, I didn't wake up early, so I'm not going to get a worm because I wasn't an early bird. I wasn't late. I'm not a late bird. Just regular time, so I'm a medium bird,” and we still use it. Thanks.
Thanks, Don. That's very logical.
If you want to share the story of your familect, a word or phrase that you only use with your friends or family, leave a message on the voicemail line at 83-321-4-GIRL or leave a voice message on WhatsApp. Be sure to call from a nice, quiet place, and if you want that number or link later, you can always find them in the show notes.
Grammar Girl is a Quick and Dirty Tips podcast, and we have lots of other shows. Tomorrow, the Nutrition Diva will have an interesting interview about how GLP-1 drugs work. So check that out if you or someone you know is using them.
Thanks to Morgan Christianson in advertising; Dan Feierabend in audio; Holly Hutchings, director of podcasts; and Nat Hoopes in Marketing, who couldn’t say his R's until he was 10.
And I'm Mignon Fogarty, better known as Grammar Girl and author of the tip-a-day book "The Grammar Daily." That's all. Thanks for listening.