poetry.DNA

Poetry.dna explores the relationship between the organic and the mechanical.It is a generative poetry bot that is modelled after the biological principles of evolution, mutation, and self-regulation. Inspired by artificial intelligences that learn based on associative phrasing (such as twitter bots) this project is given the capacity to create unique phrasing, wording, and formatting to poetry on it’s own accord. This ability is passed on through JavaScript where Poetry.dna is given a large array of data about the English language and the use of mathematics such as Markov chains (thanks to RiTa.js). What makes Poetry.dna different is it has no end goal. Most modern AI learns through a process that involves reward systems upon correct identification. Poetry.dna does not learn in the sense one might traditionally expect from an AI, instead of a static algorithm that dictates its creative direction it was instead modelled after nature. True evolution does not have an end goal, when a mutation occurs the organism either dies, or passes on their genes. AI evolution typically evolves to a specified point, when it reaches that point it is deemed successful. Poetry.dna does not have an end point, like in nature it evolves at random and does not move solely forwards but possibly backwards, it is always in a state of flux and self-regulation. It has corrective mechanisms within its coding, but it is not always guaranteed, much like genetics.

http://matthewhalpenny.com/poetry_dna/