A lot of writers, including me, ask this question because we want to create content with help from AI that doesn’t READ like it was AI-generated. Our real question is, “Will readers think my work it was written by AI, and how can I prevent that?”
Readers also want to know if something was written with AI. But they’re asking for a different reason. They want to know, “Can I trust you? Is your work worth my time and attention?” Unless told otherwise, your readers assume that they’re consuming original ideas written by an actual person. And if you break that expectation, you’ve lost them.
But…why does that happen? Let’s start by considering how you can tell something was written by AI—or if it’s even possible to tell.
How can you tell whether something was written with AI?
Apart from sloppy errors (like finding ChatGPT’s instructions in the text), it’s surprisingly tricky to tell whether something was written by AI. People THINK they can, but we oversimplify when we just look for certain patterns or phrasing.
Even I have shared some of these AI “tells” in the past:
- Em dashes (this is probably the canonical example of an AI “tell”)
- “genuinely”
- “It’s not just ____; it’s _____.”
- “…that matters most”
- Etc.
The problem is that none of those are unique to AI. So, although finding these elements (especially more than one) in a piece of writing raises the likelihood of something being AI-generated, you can’t be 100% sure.
Plus, looking for certain words and phrases is like playing AI whack-a-mole. The models are always learning and changing, so the AI “tells” list is impossible to keep current. This is the same reason AI detection tools kind of suck. They’ve gotten popular because people are desperate for a way to identify AI-generated text easily, but they’re unreliable because generative AI is always evolving.
So, you can’t look at any single element, or even a combination, or a “vibe” and know for sure whether someone was written by AI. And as Ed Gandia and others have astutely pointed out, right now is the stupidest AI will ever be. It’s only going to get better at writing as it goes along, so identifying AI-generated content is going to get more difficult, not easier.
Maybe you disagree and think, “I know AI-generated writing when I see it.” But consider this: you’re only recognizing it when it’s obvious. When it’s being used well, it doesn’t read like AI wrote it. As one Reddit user put it, “AI is like Botox. You only notice it when it’s bad.”
Don’t despair. We can’t tell for sure whether something was written by AI, but we may be asking the wrong question. It’s the reader’s perception that really matters.
Why do people get so riled up about AI-generated writing?
Some people get REALLY ticked off when they encounter AI-generated writing. Like, they’re personally offended by it. They don’t get mad that their cars are made by machines or their clothes are designed by an algorithm. But clearly AI-generated writing has hit a nerve.
To consider why, I went to the home of strong opinions: Reddit. And folks did not hold back.
The comments I encountered seemed to encapsulate some of the main reasons people are so bothered by AI-generated writing:
1. Readers assign less value to AI-written text. It’s human nature to think that low-cost, low-effort tasks are worth less.
“It feels kind of insulting that I can be reading something like a Reddit post or article, and it took me longer to read it than it took to produce it.”
“[Y]ou don’t even know if they double-checked what the AI said is actually what they think. Parts of it might be, but parts of it might not be. Why would I waste the time reading something if it doesn’t even represent what that person actually thinks?
2. AI-generated content breaches the reader’s trust because the writer has betrayed their expectations.
“There’s something about it that makes me not want to read or engage with it. It leaves me with a feeling like the person cheated, and it makes me not trust them or the post.”
3. Readers feel insulted if it seems the writer didn’t think they were worth the effort.
“[T]here’s something weirdly disrespectful about reading text that took zero real effort… the fully AI-written essays clogging feeds lately? Just exhausting.”
“When I get served up AI from someone, I feel like I received the minimum amount of effort they felt they could get away with, and that feels insulting.”
“I don’t want to read something nobody cared enough to write.”
4. A writer and reader are in an implied relationship. If the reader suspects that a text was actually written by AI, the writer hasn’t held up their end of that bargain.
“[I]t de-humanizes the conversation. I don’t want to talk to a robot, I want to talk to the person.”
“It pretty much kills the internet. No one wants to talk with an endless army of bots, forums are supposed to be used for humans to interact with one another, not with algorithms that don’t understand their own outputs.”
“It’s like going to a mediocre play performance because your friend said they were in it, but they were never in it, they never even went to rehearsal.”
5. Anyone would be angry if they felt as if they were being duped.
“[T]he intent is not clear – are you trying to pass this off as your own work and ideas? Did you even read the content that you sent me? Does it actually reflect your sentiments? Are you trying to pretend that you put a lot of thought or effort into this, when really you took a giant short cut?”
“I have ZERO interest in my money going towards AI slop of any kind, and if I think a human made something and then find out it’s AI instead I’m going to feel disgusted and angry. It’s NOT going to change my mind – it’s going to make me deeply uncomfortable.”
6. People who consider writing to be an art form are disheartened by new technology that devalues their craft.
“My question is, how do you cope with the fact that somebody may soon create in seconds what you spent a week creating, and with comparable if not better quality? How do you not get discouraged to continue writing?”
“Because most writers have spent years, if not decades, learning the intricacies of writing. It hurts to think about how much of that skill is being invalidated.”
My takeaway? Facts and information aren’t all that matter in a piece of writing. Authorship is more than just a technicality; it’s a proxy for inherently human traits like credibility, integrity, and empathy. People want to be confident they’re hearing from other people.
How can I make sure my writing doesn’t sound AI-generated?
Perhaps you think I’m building an argument against using AI in your writing at all. There are certainly hard-liners who say that writing shouldn’t be AI-generated, full stop. Problem solved. To me, it’s more nuanced than that.
Even with all of the ambiguity and concerns around AI-generated writing, I still think there’s value in using these tools. And frankly, people are going to do it anyway, even if they don’t admit it. It’s unlikely to be effective to take an “abstinence-only” stance when it comes to using AI.
I’m not suggesting hiding AI use or lying about it. This is not about getting better at fooling your reader so you can pump out more words faster. Like in most things in life, it works best if you go into it with good intentions. Don’t use AI in ways that would compromise your reputation, integrity, or originality if someone found out about it.
“Okay sure,” you say, “but I still want to know how to make sure my writing doesn’t sound like AI.” Like you, I’m experimenting with and learning new ways to do this. Here are the options I see right now:
- In some cases, don’t use AI to write at all.
I know. I’m sorry. But sometimes it’s just not appropriate. The more sensitive, personal, or relational a piece is, the more important it is to be written BY YOU. That’s not to say you can’t use AI at all. It can be a great brainstorming partner and editor. Just don’t have it write. - Use AI as a boost, but don’t let it write the draft for you.
Are you looking for a better way to phrase something or a clearer transition? Does your text need to be tightened up or edited for a particular audience? AI can help with all that stuff without writing it for you. - If you have AI do the writing, don’t outsource your critical thinking.
A lot of writing “sounds” like AI because someone accepted AI’s output without reviewing and editing it. You simply cannot check your brain at the door and expect to create something worth reading. - Use AI to write when authorship doesn’t matter.
If the goal is simply to convey information or instructions, don’t worry about it. No one cares if IKEA assembly instructions are AI-generated, as long as they work. Same with weather reports.
Because there are no universal rules governing AI and writing, it comes down to being thoughtful and responsible. Use AI in ways that engage your creativity and make your writing better, instead of trying to “get away” with using AI on the sly. As ever, having respect for your readers is the priority. They’re watching, and it matters.
Disclaimer: I often employ AI to help with my writing, but I didn’t use it here. The intent was to share my original insights and opinions, and as we know, AI can’t do that kind of heavy lifting (yet). As a result, all of the mistakes, inconsistencies, and omissions are my own.
The image, clearly, was AI-generated. But it’s cute, no?