AI as a Material for Design

Pilling, Franziska and Coulton, Paul (2023) AI as a Material for Design. PhD thesis, Lancaster University.

[thumbnail of 2023Pillingfphd]
Text (2023Pillingfphd)
2023Pillingfphd.pdf - Published Version

Download (5MB)

Abstract

From Netflix recommendations to Amazon Echos sitting proudly on kitchen countertops, artificial intelligence (AI) has been inserted into the mundane settings of our everyday lives. These ‘smart’ devices and services have given rise to the collection of data and processing within everyday objects, accumulating new challenges, particularly in legibility, agency, and negotiability of interactions. The emerging field of Human Data Interaction (HDI) recognises that these challenges go on to influence security, privacy, and accessibility protocols, while also encompassing socio-technical implications. Furthermore, these objects challenge designers’ traditional conventions of neutral interactions, which only work as instructed. However, these smart objects go beyond typical human-object relationships functioning in new and unexpected ways, creeping in function, and existing within independent and interdependent assemblages of human and non-human actants—demanding alternative considerations and design practice. This thesis aims to question the traditional practice of considering and designing for AI technology by arguing for a post-anthropocentric perspective of things with agency, by adopting the philosophical approach of Object Orientated Ontology with design research. This research ultimately presents and builds (a currently) unorthodox design approach of Human-AI Kinship that contests the design orthodoxies of human-centred design. Conclusively, this research seeks to bring into being AI as a material for design and justify through the case study of AI legibility. A More than Human Centered Design approach is established through a transdisciplinary and iterative Research through Design methodology, resulting in the design of AI iconography that attempts to communicate and signify AI’s ontology to human users. This thesis is concluded by testing the legibility of the icons themselves and discussing philosophy as an asset for design research.

Item Type:
Thesis (PhD)
Uncontrolled Keywords:
Research Output Funding/yes_externally_funded
Subjects:
?? artificial intelligencelegibilitydesign fictionphilosophyobject orientated ontologyyes - externally funded ??
ID Code:
204678
Deposited By:
Deposited On:
21 Sep 2023 08:30
Refereed?:
No
Published?:
Published
Last Modified:
26 Nov 2024 02:17