Mutating Text Toward Better Nonsense?! Recently, we’ve made a lot of progress in getting computational processes to generate text that makes sense. But creative processes (like storytelling and poetry-writing) often benefit more from injections of *good nonsense*. In this talk, I first briefly characterize what makes nonsense “good”, then discuss a series of three projects that take different computational approaches to the generation of high-quality nonsense: Blabrecs, a game that uses a statistical classifier to condition the generation of nonsense words by human players; Blabwreckage, a poetry machine that adapts the Blabrecs classifier to progressively mutate arbitrary source texts (either human-supplied or derived from random noise) into pronounceable nonsense poems; and Savelost, a poetry machine that aims to progressively compress a source text into a shorter and shorter micropoem while retaining as much of the original text’s meaning as possible. I conclude with a brief discussion of the architectural similarities and differences between these projects, and of what the similarities might imply about the relationship between computation, creativity, and nonsense. Max Kreminski is a researcher in artificial intelligence, human-computer interaction, and creativity, with a particular focus on the design, development, and evaluation of AI-based creativity support tools for storytelling, poetry, and game design. Currently they direct the storytelling lab at Midjourney; before that they were an assistant professor of computer science at Santa Clara University. Presented at !!Con 2024 (pronounced, of course, "bangbangcon"!) in Santa Cruz, CA -- https://bangbangcon.com/ !!Con is a production of the Exclamation Foundation -- https://exclamation.foundation/ Video Production by Confreaks Follow Confreaks 👇 https://confreaks.com https://x.com/confreaks
Get notified about new features and conference additions.