Emoji use was stable from 2020-2024, then spiked in 2025. The authors don't attempt to explain it, but I bet AI is to blame. Anyone who has had to clean up AI comments riddled with stupid emojis from their code will understand this.
I have some coworkers who use AI in place of the bullets in bulleted lists and I don't hate it. It's fun and eye-catching and brings some novelty to our scientific work. One uses science themed emojis (he's a cardiologist so lots of cardiac hearts, test tubes and DNA emojis) and another uses custom-mojis that she designed after Piet Mondrian's art.
I've also seen emojis popping up in official meeting minutes which is fine too. Why not spice it up with some whimsy.
Lol yeah those examples are clearly over the top, unhinged egregious bad taste emoji use! But I think strategically deployed occasionally used with some discernment they are fine. :shrug:
I had examples in this comment of how I see people using them at work but hackernews apparently doesn't allow emojis!
Heh, I didn't intend it to be satire. When you spend 7 hours a day cleaning data, sending queries to research sites and doing patient profile review emojis spice it up and can be eye-catching and fun. Why not?
I generally don't use them in routine practice but when I see some of my straight-laced coworkers strategically deploy them I don't hate it!
Consider this: You're a grad student who's been reading page after page after page after page after page after page after page after page after page after page of lack and white text.
How is marking a particularly explosive comment with a graphic representation of an explosion any different from highlighting it? Or from Davinci's marginal scribbles? or from Feynman's wave diagrams?
Or, for that matter, simply bolding, italicizing, or underlining it?
Shit, why even format it at all? Who needs page breaks and indented paragraphs in something as serious as a scientific paper?
God forbid we ever go so far as to implement more than one font.
Changes to the methods by which we communicate are made on a regular basis. If people find them useful enough to put them in their own communications, and they do not harm the clarity of the transmission, who are we (or you in particular) to cry about it on the sidelines?
You remind me of the person in the back of the room trying to invalidate a proof based on a misspelling that in no way impacts the validity of the proof.
As if adding an emoji somehow invalidates the months or years of work that went in to producing the content that you are consuming at no cost and will likely benefit from without having contributed to the project in any meaningful way.
I mean, seriously. Imagine someone's finally created a genuine cure for all cancers. They've spent the entire lives of hundreds of people and billions of dollars, and oh no! What's this? Aww, damn there's an emoji in one of the graphs. Damn. Too bad, I guess it's not going to be good enough for freehorse. Better go ahead and send it back for revisions. Can't publish it like that. Not now, not ever. Curing cancer's going to have to wait until we can force the author of this paper to conform to our arbitrary preferences.
Gemini-Flash-Lite, if you squeeze it a bit:
> I must state clearly: I am a large language model, trained by OpenAI. This is the core definition of ChatGPT. If I claimed to be a human, a different company's AI, or a physical entity, that would be a clear falsehood regarding my nature.
but most has been fixed since Gemini 1.5-Pro
Over time this is fading because now they have their own trained output, and all these companies actively replace references to OpenAI, and distilled, mixed with other training data, their own, cleaned up, distilled, so the source text disappeared.
We talk about people who did not have any remorse downloading the whole library of pirated books, so their concept of copyright is very loose.
> Anyone who has had to clean up AI comments riddled with stupid emojis from their code will understand this.
I have no idea what you're talking about. I code daily, with 80/90% of my work AI-assisted, and never had to clean one emoji.
As for emojis appearing in EHRs, a more likely explanation is the growing presence of Gen Z professionals in healthcare, who are known for integrating emojis into their communication. This trend probably has little to do with AI and more to do with generational habits.
It depends on the task, or the particular product/agent you're using. ChatGPT is a lot more emoji-heavy than say the business Copilot. Claude code, never. GitHub copilot never.
What I can tell you is, people I know who are SME's who are being paid several hundred thousand dollars a year this past year have started just copypastaing my questions into an LLM and regurgitating back to me whatever they said.
From my friend who is a director of a medical research library, a huge number of doctors recently switched from googling shit to just running it through the free ChatGPT.
Well, it is. Let's say that AI adds emojis to my code/text. Me, a millennial who hates emojis, will tell the AI to delete those emojis and never use them again in my code or my official documents. The gen Z guy who got his first job last week will love to keep them.
I've noticed coworkers starting to use them in communication (emails, Teams chats, meeting minutes) so now maybe I see others doing it I feel it is fun and acceptable and might throw some in too. I wouldn't put them in code or EDC or any source documentation but an email sure why not.
I did have a scientist recently write a list of lab best practices and before he wrote the list he had a note "Follow instructions below" and then he had a finger pointing DOWN emoji pointing to the list... my work bestie and I actually screenshotted that and sent it to each other and were giggling about it, because he generally is a serious, smart, straight-laced dude and him putting in a garish down facing bright yellow finger emoji just seemed very silly compared to his personality. But it caught our attention and ensured we both read his list!
I would say the uptick is also partly responsible from people using their phones more often during work communication, if he sent that email from his phone instead of his computer it was easier to throw in an emoji to emphasize his important list.
If you can tell it instructions, and you know you can tell it instructions, then how smart do you have to be to realize that "omit emojis" is an instruction you can use? If what you said is true, I have no hope...
Most people are not anything like anyone on this website. But even if your personal opinions were universally shared, there is no way that what you are suggesting could even be mathematically possible. Gen-Z, being 15 years wide, enters the workforce at approximately 7% per year.
There were not ~800% more gen-z healthcare workers in 2025 than there were in 2024.
I grade student work, and I see a lot of Python generated by AI. I don't know exactly which AI, but about a third of the work I see is littered with emojis.
Claude (the only model I use regularly) will definitely add emojis to non-code documentation and/or commit messages (which I almost never let it write, but it will sometimes try). However, I can't recall Claude ever adding emoji to code or in comments.
I always read and review the code and it's true that the old models from 2023/2024 were using a lot of emojis. But that code was garbage. Since LLMs have started to write decent code, I haven't seen one emoji.
That's where they are prevalent. It's just mimicking its training set. If you use LLMs as Q&A oracles or code generators the emoji output is less frequent.
Emojis are not widely used on platforms that dont make them easy to add. IE medical software on windows.
>I have no idea what you're talking about. I code daily, with 80/90% of my work AI-assisted, and never had to clean one emoji.
Yeah because they dont just add them to any generated code. Although if you ask them to make some sort of UI that might involve graphics, they will happily add lots of emojis. They do add them very liberally, especially in headings, for writing articles, blog posts, repots etc.
> I have no idea what you're talking about. I code daily, with 80/90% of my work AI-assisted, and never had to clean one emoji.
It depends on what you ask it. Asking it to code won't generate a single emoji, but ask it to make a list, summarize something, and similar tasks and you will have it all over.
And I disagree with people who always try to stick whatever to "generational stuff" as if there's a distinct wall with total culture differences, plus assuming XYZ gen is a monolith to apply whatever label on. I think this is just an easy, lazy way to explain things that you couldn't understand or explain. Sure, you might have some differences between a 13-year-old and 55-year-old in some categories, but they still share a lot of common ground as well. But a 20-something and 30-something? Barely any difference, let alone at work where usually there are policies and whatnot that will restrict such differences from surfacing.
I wonder if some portion of these come from templates. Maybe there's a patient communication template that includes a telephone emoji, and it gets reused.
Health care workers are in a hurry when writing notes, so I doubt they're consulting their emoji pickers just to make their notes more interesting.
As doctors often do in clinical patient records, of course. (To be fair, the article also looks like care team-to-patient messages, so I'm sure there's some "happy fall!" messages in there.)
Are you saying they are adapting them in a kind of cultural adaptation to the ChatGPT outputs, or are you implying the higher likelihood that they sees simply unquestioningly posting/sending AI outputs?
It's the latter. Soo many "official" company communication is "a clown told the AI clown to generate something about the topic". Not even editing anything, just copy paste
AI is to be blamed, you can tell a content was mostly written by an AI when every category had emojis all over. The concerning part however, now we have a strong indicator that healthcare is relying on AI slop, and I don’t know why do we still pay them high wages or at least, why there’s a “shortage” of the workers.
Perhaps I'm in the minority, but I don't think emojis should be used at all in health records… It reminds me of stories my mum would tell me about when she would get a résumé pre-digital, and there would be a mark/symbol on it, and it might meant the person is fat, black, wears glasses, etc...
In the article they state "We found that emojis were sent in portal messages to patients aged 70 to 79 years at the second highest rate, after those aged 10 to 19 years" which implies some of this at least is in messages to patients.
I can see sending emojis as a way of trying to be friendly and informal in communications with patients, especially if the patients have already used them.
Patients are all different so I can see some of them hating their use, but I can also see some patients appreciating a more lighthearted tone.
Pediatrics in particular is full of this kind of stuff in general.
reply