You used to end every message with a smiley face. A thumbs up meant "got it." A string of fire emojis was the highest compliment you could pay someone's selfie. Emojis were playful, human, spontaneous. Then AI started using them too. And now every time you drop a 😊 into a work email, there is a small voice in the back of your head asking: does this make me look like a chatbot? The story of emojis is one of the most interesting arcs in digital communication. They were born out of necessity, grew into a global language, became culturally contested across generations, and are now caught in the crossfire of the AI authenticity crisis. What was once the most human thing about digital text is starting to feel like the least.

How emojis began

The first emojis appeared earlier than most people think. In 1997, Japanese mobile carrier SoftBank (then J-Phone) released the SkyWalker DP-211SW phone with a set of 90 distinct emoji characters, including, notably, the poo emoji. But the more famous origin story belongs to Shigetaka Kurita, a graphic artist working at NTT DoCoMo. In 1999, Kurita designed 176 tiny 12-by-12 pixel images for i-mode, DoCoMo's mobile internet platform. He wanted a way to convey information quickly and visually, like showing the weather forecast with an icon instead of spelling out "cloudy." Kurita's original set was mostly symbols, not faces. Arrows, weather icons, sports, phases of the moon. They were practical, not emotional. His 176 designs are now part of the permanent collection at New York's Museum of Modern Art. But even Kurita's set was not truly the first. Research by Matt Grogan uncovered that Sharp's PI-4000 personal organizer, released in 1994, included recognizable emoji characters in its text input system. The history of emoji keeps getting pushed further back as people dig through old Japanese consumer electronics. Before any of these, there were emoticons. Puck Magazine published "typographical art" in 1881, showing joy, melancholy, indifference, and astonishment rendered in type. Scott Fahlman proposed :-) and :-( for online message boards in 1982. The Japanese kaomoji tradition, with faces like (╥_╥) and (◕‿◕), developed its own rich vocabulary through the 1980s and 1990s. Emojis did not appear out of nowhere. They were the visual culmination of over a century of humans trying to put feelings into text.

The golden age

Emojis went global when Unicode Consortium standardized them. In October 2010, Unicode 6.0 encoded hundreds of emoji, and suddenly these little pictograms could travel across platforms and devices. Apple had already created its own emoji keyboard for the Japanese iPhone market in 2008, borrowing heavily from SoftBank's designs. When that keyboard became available worldwide, emoji use exploded. By 2015, the 😂 Face with Tears of Joy emoji was named Oxford Dictionaries' Word of the Year. That moment marked the cultural peak. Emojis were everywhere: in marketing campaigns, in court cases, in academic papers about digital linguistics. Linguist Vyvyan Evans argued in his book The Emoji Code that emoji was the next step in the evolution of human language, a visual form of communication that was "both resonant and powerful." The numbers kept climbing. Unicode kept adding new emoji with every release. By 2022, there were 3,633 emoji. By 2025, the count reached 3,953. New batches included skin tone modifiers, gender variants, flags, and increasingly niche symbols. The emoji lexicon was growing to accommodate the full complexity of human identity and culture. But somewhere in that expansion, something started to shift.

The generational fracture

Gen Z did not reject emojis. They remixed them. And in doing so, they broke the shared understanding that made emojis work as a universal language. The most visible example is the thumbs up emoji 👍. For millennials, boomers, and Gen X, it means "okay," "sounds good," "acknowledged." For Gen Z, it often reads as passive-aggressive, dismissive, or cold. Psychology Today documented the divide, noting that younger users on Reddit described receiving a thumbs up as feeling like "walking into your boss's office, saying something important, and watching them look you in the eye, give you a slow thumbs up, and turn back to their screen." The 🙂 slightly smiling face underwent a similar transformation. Older users see it as friendly. Younger users see it as deeply sarcastic, the emoji equivalent of saying "fine" through clenched teeth. The 😂 face with tears of joy, once the most popular emoji in the world, became a marker of being "old" online. Gen Z replaced it with 💀, meaning "I'm dead" (from laughing), a usage that bewilders anyone over 35. A study published in the Advance Social Science Archive Journal found that Gen Z assigns different meanings to emojis than millennials, who take them more at face value, while baby boomers use them most literally. Research from the Adaptivist Group showed that 44% of Gen Z workers prefer to use emojis ironically, compared to just 17% of millennials and 12% of Gen Xers. Adobe's data found that 74% of Gen Z uses emojis differently than their intended meanings. This was not a crisis yet. Generational slang has always existed. But it meant that emojis were no longer a universal shorthand. The same symbol could mean opposite things depending on who sent it. The shared visual language was becoming fragmented.

Then AI showed up

AI did not kill emojis. But it contaminated them. The problem is simple: large language models love emojis. When ChatGPT, Gemini, or any other AI assistant generates text, it reaches for emojis with an enthusiasm that no human would naturally match. Customer service bots pepper their responses with 😊 and ✨. AI-generated social media posts stack emojis at the end of every sentence. Marketing copy produced by AI tools arrives dripping with 🚀🔥💡. The sparkles emoji ✨ became so synonymous with AI features that The Wall Street Journal ran a piece about how it became "the symbol of our AI future." Every major tech company, from Google to Apple to OpenAI, adopted it as the icon for AI-powered features. A symbol that once meant magic or excitement now means "this was generated by a machine." Netcraft's cybersecurity research identified excessive emoji use as a reliable indicator of AI-generated content, particularly in phishing sites and automated scam messages. The pattern is consistent: AI overuses emojis because they are safe, positive, and high-engagement tokens in the training data. The model reaches for them the way it reaches for em dashes, not because it understands the emotional context, but because statistically they appear frequently in the kind of text it was trained on. This created the same dynamic that turned the em dash into a "ChatGPT tell." Once people noticed the pattern, the association stuck. And now humans who use emojis naturally find themselves under suspicion.

The robotic feeling

A Facebook user in the NBCTRIBE group captured the shift perfectly: "I used to love adding emojis to my emails, and my posts. But I started having issues with it when people accustomed my emails to AI written just because of emojis. I felt uncomfortable with it, so I stopped adding emojis." This is the same self-censorship pattern that happened with em dashes, with exclamation points, with any linguistic feature that AI happens to overuse. People do not stop using these tools because they have become less useful. They stop because the social meaning has changed. An emoji that once said "I'm being friendly" now says "this might be automated." The irony is painful. Emojis were invented specifically to add human warmth to digital text. Kurita created them because typed messages felt cold and ambiguous without visual cues. They were the antidote to robotic communication. Now they are the symptom. In professional settings, the effect is amplified. A Preply survey found that roughly 80% of U.S. adults have been confused by someone else's emoji use. When you add the AI suspicion layer on top of existing generational misunderstandings, emojis in work emails become a minefield. Use too many, and you look like a bot. Use the wrong ones, and you look out of touch. Use none, and you sound cold.

Emojis are not dying, but they are changing

It would be easy to declare emojis dead, but the data does not support that. Around 92% of people still use emojis. Buffer's analysis of social media posts in 2025 found that the sparkles emoji ✨ maintained the number one position all year, followed by hearts, check marks, and fire. Emoji use is not declining in volume. What is changing is the relationship people have with them. Emojis are becoming more self-conscious. Every emoji choice now carries more weight, more potential for misinterpretation. The spontaneity is draining out. You used to throw a 😂 at the end of a message without thinking. Now you think. And the thinking itself changes the nature of the communication. The Unicode Consortium keeps releasing new emoji, but the batches are getting smaller. In 2022, there were 112 new emoji. In 2023, just 31. In 2025, an all-time low of 8. The emoji vocabulary is still growing, but the pace of expansion is slowing. The frontier is no longer about adding more symbols. It is about figuring out what the existing ones even mean anymore.

The deeper pattern

Emojis are just the latest casualty in a broader phenomenon: AI is making every form of digital expression feel suspect. Em dashes, exclamation points, certain turns of phrase, emojis, even the absence of typos, all of these have become potential markers of artificial text. The result is a strange new form of communication anxiety where humans modify their natural writing habits to avoid being mistaken for machines. This is not a technology problem. It is a trust problem. When you cannot tell whether a message was written by a person or generated by a model, every signal gets scrutinized. Emojis were always a shortcut for sincerity. A quick way to say "I mean this warmly" or "don't take this too seriously." But shortcuts only work when everyone agrees on what they mean. Right now, nobody agrees. The fall of emojis is not really about emojis. It is about the slow erosion of the signals we use to recognize each other as human in digital space. Every time AI adopts a human communication pattern, that pattern loses some of its power to convey humanness. We keep retreating to new signals, new markers of authenticity, and AI keeps following. Kurita designed those first 176 tiny pixel images to make mobile internet feel less mechanical. Twenty-seven years later, the most mechanical things on the internet use his invention more enthusiastically than anyone else. The emojis did not fall. They were picked up by the wrong hands.

References