Towards Human-Centred AI-Co-Creation: A Three-Level Framework for Effective Collaboration between Human and AI
Year: 2023
Type: conference-paper
Abstract: AI-based creativity-support systems are gaining attention from designers and researchers. However, a research gap exists on how to tailor those systems by maximizing flexibility based on human needs and preferences. This study proposes a schematic human-AI co-creation framework to maximize system flexibility and enhance creative outcome generation. The framework proposes the involvement of AI in three levels of creation and allows humans to adjust between the levels anytime during the creative process based on their preferences. We tentatively define how AI should collaborate at the three levels. To implement the framework, a co-creation system (GSM) was built to support humans in creating sculpture maquettes with AI. It includes three key components: a prompt-based generated model (DALL·E), advanced computer vision, and robotic arms. A user interface is provided to ensure transparency. Preliminary user studies have demonstrated that the system enhances flexibility and allows users to generate more creative maquettes.
Authors: Mingyuan Zhang, Zhaolin Cheng, Sheung Ting Ramona Shiu, Jiacheng Liang, Cong Fang, Zhengtao Ma, Le Fang, Stephen Jia Wang
Institution: Hong Kong Polytechnic University
Language: English
Topic: Digital Media and Visual Art
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
Funder: Hong Kong Polytechnic University
Awards: G135006850, G6960633787
OpenAlex ID: W4387606666
