Interactive evolution of artificial life art
Abstract
In this thesis, we designed and presented an interface which is used for creating art using tools from artificial intelligence and artificial life. The interface is used for conducting two different experiments, one for the control group and one for the test group. It is tested and validated among our 34 participants who did both the said two experiments and a short online survey after each experiment with the aim of investigating how AI algorithms can be used for enabling all individuals even with disabilities or impairments, to express their artistic creativity. The interface uses evolutionary algorithms in generating animated images and short clips of life-like patterns of pixels which are based on MergeLife update rules describing local updates for cellular automata (CA). Participants are asked to select a sequence of rules representing various images which appear the most appealing to them, either randomly generated (control group) or assisted by an interactive evolutionary algorithm which considers the participant’s previous choices (test group) within the 10 sets of image selection. The results have shown a significantly higher satisfaction of the test group with a significance level of 95% and a power of 82%. Finally, we also had a discussion for future directions of the said study and talked about the usefulness of such an interface for people with disabilities and limitations. I wrote a research paper together with my thesis supervisors called An HCI experiment to explore interactive artificial life art (Dumo, Lind, & Nichele, An HCI Experiment to Explore Interactive Artificial Life Art, 2022) which is based on this thesis. It has been accepted for publication and presentation in the 24th International Conference on Human-Computer Interaction which will be held virtually from 26 June to 1 July 2022.