She doesn’t exist. Yet she’s everywhere—and she’s all of us.

“Lately, Elara had begun to feel like the only thing she truly knew about herself was what she wasn’t. She wasn’t a graphic designer, wasn’t a kindergarten teacher, wasn’t even a particularly enthusiastic dog walker anymore….”

Elara also wasn’t real, on any level. An artificial intelligence chatbot created the character in response to my prompt: “Can you write a brief start to a story about a woman’s journey of self-discovery?” If you couldn’t guess how synthetic Elara was from her unlikely trio of careers, her name would still be a dead giveaway.

Elara is an extremely rare name in the human world. In the world of generative AI, though—the systems designed to generate their own new content like text, images, and videos—the name is a smash hit. Elaras abound in the output of every generative AI system, from all-purpose powerhouses like ChatGPT and Gemini to niche character generators. They populate every writing genre from math word problems to novels. The name has become an embodiment of the low-effort machine-generated content known as “AI slop.”

Elara is a product of this moment, both in technology and names. It’s a point of convergence that challenges our notions of style and creativity. And as AI takes deeper root in all corners of our lives, it may be a glimpse into the future. That makes Elara the 2025 Name of the Year.

Meet Elara. Frequently.

If you haven’t encountered Elara before, let me introduce her through the words of the people who know her best: the writers, gamers, AI aficionados and even teachers who are steeped in machine-generated writing. Throughout the internet they ask each other, why are the chatbots obsessed with the name Elara?

“Every single time I have a chat with a bot and there’s a female character introduced by the bot, their name is going to be Elara. Doesn’t matter the location, the time period, the setting—it is always Elara.”

“It’s kinda messing with the immersion. At some point, it’s more about waiting for something or someone named Elara to show up than enjoying the story itself.”

“I teach high school English….I gave a creative writing assignment: write any story you want. In the fine print, I added one rule — if your main character’s name is Elara, you lose 99 points.”

“It should be called The Elara Effect.”  

AIs just can’t stop giving characters the same names, over and over. Elara is the epicenter of the phenomenon, but it spills over into phonetically similar names. The classics Elena and Clara are also overrepresented, and the elaborate Eldoria pops up in fantasy and science fiction settings. When a surname is required for these characters, the AIs favor Voss or Vance. One frustrated AI-assisted writer noted, “I actually have only one author’s note running constantly: ‘Do NOT ever generate the name Elara, or the surname Vance.’”

Chatbots’ output has become far more sophisticated in the past year, and Elara has started to break containment. On Goodreads, a list of AI-generated books featuring characters named Elara includes 120 titles. In August, the Read Max blog reported that 62 books for sale on Amazon were credited to authors named “Elara Voss”: AI-generated authors for AI-generated writing. The number of “Elara Voss” books has grown in the months since, including a dystopian novel aptly titled 404 Humanity Not Found.

I decided to test out the name phenomenon for myself. I asked three major AI chatbots to start stories about women, with five different bare-bones genre prompts: suspense, romance, fantasy, journey of self-discovery, and simply “a woman walking in a field.” The 15 character names that resulted:

Elena
Elena (yes, again)
Lena
Eleanor
Evelyn
Clara
Clara
Sarenya
Isabella
Grace
Sarah
Mara
Maya
Lyra
…and Elara.

Clearly, the AIs have a type. It’s not a type tailored to any particular setting or character trait, since it arises context-free across genres. Nor is it popularity-based. The names range from current top-10 hits to virtual unknowns. As for Elara specifically, you can dig up prior examples of the name, including minor characters in Greek mythology and video games. That’s true of countless names, though, and it doesn’t account for the Elara-Clara-Elena nexus.

So where did the AIs get the idea to make such an uncommon name their paragon? The same place they get most of their ideas: from us.

Modeling all-too-human intelligence

We train AI language models to reason and write by feeding them vast quantities of human-generated information. They read millions of books, newspapers, Reddit threads, ads, Wikipedia articles and online arguments, then aggregate and average the information. Their realm of knowledge is immense, but they have their weaknesses and biases—as do we. They train on human data, so they absorb the tendencies in our own behavior.

The images of “a woman named Elara” that started this article are a good example. Think about the pictures of people you see in media each day. Not only Hollywood stars, but the hundreds of images that pass by almost unnoticed, like store signage and junk-mail catalogues. You see a lot of pretty young white women with long, straight hair. If you narrow it down to the images intended to look attractive or aspirational, the percentage is even higher. AIs see all those images, too.

Then consider that chatbots are tuned to give results that please us, even to the point of flattery. (Me: “Can you find the median of this data set?” AI: “This is a great application of statistics!”) Is it any wonder that when I asked them to generate a picture of a woman, they defaulted to a stereotypical beauty standard? They produced what we, as a society, have shown we want.

We’ve shown what we want in modern baby names, too. We want smooth. We prioritize vowels over consonants, and limit the impact of the consonants we do use. We keep them from gathering together in dense clusters (a la Herbert and Gertrude) and gravitate toward gentle sounds like L, M, N and R that don’t disrupt the vowel flow (Liam, Ariana).

Each of our name choices is individual, and few parents even think about vowels and consonants when they’re choosing names. But when you look at millions of examples, as AIs do, a clear pattern emerges. For girls’ names in particular, parents have produced a generational style I call liquid names: “fluid and sinuous, with no bumps, stops or hisses obstructing the flow of sound.”

The most common first initial for an American girl’s name today is A, with E in second place. 39% of girls receive a name ending in -A, and the letter strings LA and LIA are at historical peaks.

Ten years ago, I set out to create the statistically ideal girl’s name based on letter construction. The result was “Arilia.” This spring, I reported that most of the fastest-rising girls’ names of 2024 sounded remarkably similar: Ailany, Alanna, Eliana, Aylani, Lainey, Wrenlee, Analeia, Alana.

A Name for Everyone

If I wanted to choose a universally pleasing female character name in 2025, the “liquid name” recipe would be my starting point. I’d bar any consonants except L, M, N and R, and make sure they’re separated by vowels. I’d aim for average length (five to six letters) and straightforward spelling and pronunciation. To maximize the breadth of appeal, I would also aim for cultural neutrality. And of course, the name would have to end in -A.

The data-aggregating AIs don’t think the same way I do. Yet the names they keep returning to check all my boxes. Which is to say, they’re doing an admirable job of modeling us as we pursue an amorphous target of desirability.

Among traditional names, Elena is an excellent fit. It’s simple, composed of popular liquid ingredients, and used in many different cultures so it’s not tied to one image. But for the purest neutrality of appeal, the best choice might be a name with no built-in associations at all. A name that few of us have any prior experience with, despite its attractive sound. Elara is just about perfect.

In fact, Elara may be a little too perfect. It approaches the realm of archetype instead of individual, much like the AI images of women created to match the name.

AIs now allow each of us to become a full-fledged culture factory, stamping out a steady stream of polished words, pictures and music. But the uniformity of mass production makes us appreciate the unique irregularities of artisan crafts. Idiosyncrasy lends personality. Perhaps that’s the downfall of a universal name; you have to be willing to turn off some people to fully captivate others.

Names then, now and next

In 2011, as companies began to name products that functioned like people, I declared Siri the Name of the Year. Fourteen years later, it’s the products that are naming the humans. We’re experiencing their “taste,” on repeat.

Could a phenomenon like Elara have happened in past name-style eras? For most of English name history, parents selected from a small, widely accepted set of “normal” names. Everyone heard names like John, William, Robert, George, Mary, Elizabeth, Katherine and Anne everywhere they went. In that kind of naming culture, it’s hard to imagine a new neutral name. Novelty itself was anything but neutral.

What’s more, the classic names all sounded different from one another, so averaging would have been problematic. The most universally acceptable name might simply have been the most common. And if some magical 18th-century AI used names like William and Mary a lot, who would even notice? That was the norm.

Over the past two generations we have overthrown this naming regime. Names are now objects of fashion with unlimited creativity, and the classic names that dominated for centuries are vanishing. In place of tradition, sound has become the dominant force in name popularity. It’s subtle but pervasive, and cuts across many social boundaries. The parents naming their daughters Lena, Aaliyah and Lainey might have different backgrounds and different priorities, but sound leads them all in a similar direction. That opens the possibility of AI language models constructing a brand-new name that fits in everywhere.

Now we’re entering a new era, one in which the language models aren’t just training on human content but producing vast amounts of content of their own. Machine-generated writing floods the web, and authors like “Elara Voss” are producing books at a growing pace. The AIs will continue to absorb all that new AI-generated material, including the many, many Elaras. Parents will also turn to AI chatbots for baby-naming advice, and I’m pretty sure I know what the chatbots will consider a good, safe suggestion.

But hey, at least Elara is a nice-sounding name. I think so. Everybody thinks so. Don’t you?

---

My thanks to everyone who participated in the NOTY discussion, to my family for being my sounding board, and to every reader who stuck with Namerology during this challenging year. Here’s to a good and name-filled 2026 for all of us!