Embodying the Glitch Perspectives on Generative AI in Dance Practice

2025-05-02 0 0 2.52MB 5 页 10玖币

侵权投诉

Embodying the Glitch: Perspectives on Generative AI in Dance

Practice

BENEDIKTE WALLACE, University of Oslo, Norway

CHARLES PATRICK MARTIN, Australian National University, Australia

Fig. 1. A dancer and choreographer exploring the glitch

What role does the break from realism play in the potential for generative articial intelligence as a creative tool? Through

exploration of glitch, we examine the prospective value of these artefacts in creative practice. This paper describes ndings

from an exploration of AI-generated "mistakes" when using movement produced by a generative deep learning model as an

inspiration source in dance composition.

Additional Key Words and Phrases: dance, generative deep neural networks, glitch

1 MOVEMENT INSPIRATION FROM AI

In our work with generative AI models of human movement, we have come across many examples of models

failing to produce movements that are human-like. We refer to these "failings" as glitches. These glitches can take

many dierent forms. Sometimes the models get stuck in a loop of movement that it keeps repeating, other times

the output becomes completely unrecognizable as a human form. Occasionally the output falls in between noise

and repetition, slipping into chaos and re-emerging from it.

In previous work, we trained several generative deep neural networks on a data set of 3D motion capture data

of improvised dance. Our implementation does not include any inbuilt movement constraints, instead, the models

need to learn these constraints from the underlying training data. As there are also no corrections applied to the

generated output, the models will occasionally produce impossible movements. Limbs can extend and rotate

past human limits at velocities that would break physical laws. We often include examples of these glitches in

presentations, both to show how the model improves during training and how dierent sampling strategies aect

the types of glitches that are produced [7]. Interestingly, the audiences that are most drawn to these "mistakes"

Licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). Copyright remains with the author(s).

GenAICHI: CHI 2022 Workshop on Generative AI and HCI 1

arXiv:2210.09291v1 [cs.HC] 5 Oct 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

EmbodyingtheGlitch:PerspectivesonGenerativeAIinDancePracticeBENEDIKTEWALLACE,UniversityofOslo,NorwayCHARLESPATRICKMARTIN,AustralianNationalUniversity,AustraliaFig.1.AdancerandchoreographerexploringtheglitchWhatroledoesthebreakfromrealismplayinthepotentialforgenerativeartificialintelligenceasacreativ...

展开>> 收起<<

Embodying the Glitch Perspectives on Generative AI in Dance Practice.pdf

共5页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Embodying the Glitch Perspectives on Generative AI in Dance Practice

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: