By “recognizing” things and patterns that were not given, inceptionist neural networks eventually end up effectively identifying a new totality of aesthetic and social relations.
But inceptionism is not just a digital hallucination. It is a document of an era that trains smartphones to identify kittens, thus hardwiring truly terrifying jargons of cutesy into the means of production. It demonstrates a version of corporate animism in which commodities are not only fetishes but morph into franchised chimeras.
— Hito Steyerl, Duty Free Art
Multiverse is a proof-of-concept prototype of a system for cybertextual generative literature.
Multiverse affords cybertextual user-readers a dialog with text generating neural networks, together exploring literary space. It is very much influenced by this Jean-Pierre Balpe’s essay Principles and Processes of Generative Literature. This exploration is bidirectional is nature, and is supported by an interactive spatial map, providing direct feedback and scrubbing of the currently explored literary space.
The current version uses GPT-3 to provide completions for an SPA front-end. If you don’t have an API-key or prefer running the models locally, see commit 32b398118ccd8e0c6807a2723dad7c7b7c44d686 for the older client-server version.
To build the application, run $ make release. The resulting bundle is found in resources/public/js/ for hosting. Alternatively, you can run $ make dev to get the self-hosted development environment.
Due to licensing, this repo does not include the typefaces used in the original project. If you are so inclined, they might be found via the marvelous Future Fonts: Obviously and Pigeonette.
This prototype was part of my master thesis in interaction design at Malmö University. To cite it, use the following biblatex entry:
@masterthesis{lagerkvist2020multiverse,
author = {Love Lagerkvist},
title = {Neural Novelty: How Machine Learning Does Interactive Generative Literature},
school = {Malmö University},
year = 2020,
}That said, you probably don’t have time to read 60 pages. Instead, you can read the IoT ‘20 paper I wrote about it with Maliheh Ghajargar:
@inproceedings{10.1145/3423423.3423424,
author = {Lagerkvist, Love and Ghajargar, Maliheh},
title = {Multiverse: Exploring Human Machine Learning Interaction Through Cybertextual Generative Literature},
year = {2020},
isbn = {9781450388207},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3423423.3423424},
doi = {10.1145/3423423.3423424},
booktitle = {10th International Conference on the Internet of Things Companion},
articleno = {1},
numpages = {6},
keywords = {interactive machine learning, interactive literature, machine learning, generative literature, cybertext},
location = {Malm\"{o}, Sweden},
series = {IoT '20 Companion}
}