Learning to be Human

Hazel Smith

Research output: Creative WorksTextual Works

Abstract

The screenshot on the title page is from The Character Thinks Ahead by Hazel Smith and Roger Dean, a piece that employs computational text generation. A video of one version of the piece can be viewed at http://www.australysis.com/hear-seeread/ aLYS-works/hearsr-audio/CharacterThinks-PresentVideoEdv2-Cp2- AudioCp.mp4 ‘Trilling Feathers’ and ‘Pruned Time Bears Anything’ are both computer-assisted poems. These poems are based on text generation performed by Roger Dean in the computer platform Python using deep learning (machine learning) techniques. The computational outputs are then shaped by me through various forms of poetic system or procedure. In both cases my intervention is quite substantial. ‘Brewed, But’ and ‘Fame’s Hark’ are computer-generated poems that appear largely in the same form in which they originated. Again, these poems arise from text generation in the computer platform Python, performed by Roger using deep learning (machine learning) techniques. However, they have not been markedly shaped by me — though the use of bold highlighting is mine in ‘Brewed, But’. The other poems are not computer-assisted or computer-generated. ‘James Comey and Donald Trump: A Revenge Drama’ draws on a number of media interviews with James Comey, ex director of the FBI, about his book A Higher Loyalty: Truth, Lies, and Leadership, Flatiron Books, 2018. In ‘Seeing Red’ the material in the left-hand column was taken from several internet sources using Google searches. The sources include Fortis Healthcare, Cosmopolitan, The Guardian, Gizmodo, HuffPost and Popular Science. ‘Personhood: a few preliminaries’ is the only poem here that is written freestyle.
Original languageEnglish
Place of PublicationOnline
PublisherSOd press
Publication statusPublished - 2020

Keywords

  • Australian poetry

Fingerprint

Dive into the research topics of 'Learning to be Human'. Together they form a unique fingerprint.

Cite this