Talk: AI and Text Generation - Computer Arts Society

26^th April 2023, 19:00
Zoom

Geoff Davis surveys the rapid rise of AI text generators and their creative implications for writing. Tracing GPT-2 to GPT-4, he explores how writers use large language models for drafting, editing and adaptive storytelling, and introduces four anthology contributors who approach human-machine collaboration from different creative, academic and educational perspectives.

AI Generated Summary

Video Recording

Geoff Davis introduces AI text generation as a creative resource, outlining its development from rule based systems to language models and considering authorship, training data, employment and claims about intelligence. Examples show text functioning as literature, visual material and a partner in artistic experimentation. Tivon Rice presents Models for Environmental Literacy, combining field research, digital environments and dialogues between models trained on ecological philosophy, fiction and scientific reports. He emphasises understanding datasets, constructing productive prompts and editing outputs rather than surrendering artistic judgement. Ray LC compares his own variations on poems with curated GPT-2 versions, exploring imitation, expressiveness and readers' perceptions of authorship. Related projects connect generated language with climate action and interactive play. Maria Cecilia Reyes describes resonant experiments with poems and images, arguing that people supply meaning to computational outputs. She considers immersive narratives that respond to choices or bodily signals, while identifying coherence and dramatic progression as continuing challenges. Shu Wan discusses classroom exercises that encourage critical engagement rather than prohibition, protecting student privacy and requiring acknowledgement of machine contributions. Questions examine prompting, stylistic guidance, accessible interfaces and differences between older, unpredictable models and newer systems. Across the presentations, co-authorship involves selection, interpretation, responsibility and human agency.

AI Generated Summary

About the Event

In 'AI and Text Generation', Geoff Davis, an AI and tech researcher at the UAL Creative Computing Institute, surveys the rapid rise of AI text generators and their creative implications for contemporary writing. Introduced by Sean Clark, Davis traces the evolution from early tools such as Talk to Transformer and GPT-2 to today's large language models including GPT-4, and explains how writers use them for brainstorming, drafting, editing, adaptive storytelling and cross-disciplinary art. He illustrates the trajectory with his own early procedural story generator, the AI Creative Writing Anthology he edits, and works by artists including Mark Webster and Sasha Styles. Davis also summarises the history of rule-based expert systems, neural networks and generative pre-trained transformers, then examines ethical concerns around the sourcing of training data, copyright, employment and academic integrity, and touches on the debate over artificial general intelligence. Hosted by the Computer Arts Society at the BCS in April 2023, the talk introduces four speakers from the anthology—Tiffan Rice on environmental literacy, Ray LC on poetic reformulations, Maria Cecilia Reyes on AI as co-writer for fiction and poetry, and Shuwan on teaching with ChatGPT—each presenting a different view of human-machine collaboration in creative art, academic research and education.

About the Speakers

Geoff Davis

CASCOM, Exhibitors, Speaker

Geoff Davis is a pioneering generative artist, writer, musician and AI researcher whose practice explores text, technology and creativity. He founded Micro Arts in London in 1984, creating and distributing computer art and story generators on data cassettes and Prestel Teletext network. His work has been exhibited internationally since 1985 and is held in collections including the Computer Arts Archive, Francisco Carolinum Museum, Franke Foundation, The National Museum of Computing and Le Random. A PEN-published fiction writer, he has also produced creativity software, worked in music, taught at universities and art colleges, and researched interfaces, AI and electronic art. He serves on the Computer Arts Society committee.

https://geoffdavis.art

Tivon Rice

Exhibitors

Tivon Rice is an artist and educator based in Seattle whose work explores digital culture, technology, representation and communication through film, installation, sculpture and AI-generated narratives. He holds a PhD in Digital Art and Experimental Media from the University of Washington and was a Fulbright scholar in Korea in 2012. Rice was among the first artists to collaborate with Google Artists + Machine Intelligence and is an affiliate of the University of Washington eScience Institute. His work has been exhibited internationally in New York, Los Angeles, Seoul, Taipei, Amsterdam, London, Berlin and São Paulo. His projects often combine emerging technologies with photography, cinema and physical materials.

https://www.tivonrice.com/