For this course, algorithmic text generation refers to the automatic creation of texts using algorithms1, or rules. These rules or conditions can be stated as “replace every noun in your pre-written poem with the last noun you used in a conversation” and performed by humans. There are also machine learning algorithms that use language models2 as training data to generate new texts that mirror human language.
This course grew from my interests in Oulipo (and other constrained writing techniques) and “programmatic situations structured by artists”3 such as prompts, instructions, choreographies, codes, or constraints used to invite others to create text content. These algorithms are often used to experiment with form and content to generate writing that is both aleatoric and unexpected. In recent years, machine learning algorithms have been developed to predict and mimic human language.
We will explore text generation by reading algorithm generated texts, playing with text generation tools, and discussing the implication of algorithm text generation within the context of agency, alienation, appropriation, archiving, and authorship. We will consider the relationship between the “author,” “computer,” and “reader”4 as we reflect on what counts as a “text,” and what counts as “writing.”
Something went wrong trying to save https://www.vox.com/science-and-health/2019/3/22/18236125/apollo-moon-poop-mars-science.
Something went wrong trying to save https://www.washingtonpost.com/nation/2018/10/26/year-old-developed-code-ai-portrait-that-sold-christies/.
Oulipian games do not produce good poetry or good architecture any more than poetic formalism or proportional systems did, but that they are teases for an already inventive mind and they offer a structure to challenge and thereby sharpen an already clear intention. They are senseless things, puzzles, exercises, a workshop to open up potential in design that perhaps, in the right hands, might lead to something truly stunning.