I trained an AI model on a list of American city names and asked it to generate some new ones for me. I used this dataset and trained the model on RunwayML. These are my favorites in no particular order: Isle of Sand Inezville Town of Shire What Mailman Older Dennis Dickitties Ruby Snicket Vermillionaire Chunky Horseman Rights and Stripes Oil Shores Inequality Run-ins Trialsville Holiday Inn Inlet

I trained an AI model on thousands of band names, expecting the model to generate some new ones for me. The results weren’t at all what I expected. Using this dataset of Billboard Hot 100 songs from the last 60+ years, I trained the model on artist names alone, using…

I trained an AI model on Billboard’s Hot 100 songs from the 2000's (2000–2021, about 8k songs) and asked the model to generate some new song names for me. I used this dataset and this GPT-2 notebook to generate the song names. …

I trained an AI model on 16,000 headlines published on NYTimes.com in 2020 and asked the model to make me some new ones. What came back was a mix of what you might expect from the New York Times in a post-2020 world and what you might not. Just like…

This spring I hired a team of freelance copywriters from the 19th century. All of them are dead, but their writing survives. So I used their books as source material and trained a computer to write like 4 Victorian-era authors from the worlds of architecture, interior design and adventure novels…