All children are special snowflakes, so shouldn’t they have names that are just as unique? Welp, there’s a neural network for that.
Built by Nate Parrott, a developer and designer who’s a student at Brown University, this algorithm can generate new names and combine existing names to create unfamiliar combinations that sound like something out of a William Gibson novel.
Using 7,500 popular American baby names, Parrott trained a neural network to convert each name into numbers–a tactic often used in machine learning. By analyzing the similar features of names, it could tell which words were “name-like” and which weren’t: the algorithm could tell that “winter” and “July” are more like names than “automobile.” Over time, the network was able to organically combine common elements of names or name-like words to create new names.
Today, there’s a cottage industry around high-tech baby naming, with “smart” name generators that can conjure names similar to ones you already like, dedicated baby-naming consultants, and even a name generator for hipster parents. Want to make sure your offspring’s name is available as a URL for his future portfolio website? There’s a generator for that too.
So goodbye to the Emmas, Jameses, and Mias of 2017–and hello to Aloora, Halden, and Nitnis. Here’s Parrott’s list of 30 original names:
Some are obviously better than others, but who knows what naming conventions will be like 100 years from now? It’s good to remember that a century and a half ago, silly-sounding names like Neppie, Effa, and Icy weren’t unheard of. Maybe by 2100 it’ll be cool to have “P” as your name.KS