Session: DTM-02-01 - Discussions on Design Creativity & Ideation

Paper Number: 87505

87505 - A Study of the Exploratory Creativity Performance Between Machine and Human Designers 

Exploratory creativity (E-creativity) is used to represent the creative performance behind the exploration process when establishing conceptual space or style. Researchers have attempted to build computational E-creativity models to help human generate more creative ideas or solutions. This trend sparks the discussion on whether the performance of machine can achieve a similar level to human beings. However, the performance gap of E-creativity between human beings and machine has not been fully studied. This study aims to investigate the E-creativity performance differences between machine and human designers. To be specific, a state-of-the-art model DALL-E is chosen as a representative of machines for generating E-creativity imagery and is compared to novice designers who are the representative for generating E-creativity imagery of humans. Expert designers are recruited as assessors to assess the creativity and E-creativity performance of the collected human and machine data. The experimental results reveal that the creativity level of humans is higher than that of machine. The E-creativity level of machine is higher than that of humans. The textual E-creativity performance is higher than the imagery E-creativity performance of humans while it is lower than the imagery E-creativity performance of machine. The results provide insights for supporting the development of more advanced E-creativity engines and corresponding evaluation methods.

Presenting Author: Liuqing Chen Zhejiang University

Presenting Author Biography: Dr. Liuqing Chen is the ZJU-100 Young Professor (and a doctoral supervisor in computer science, design science) at the College of Computer Science and Technology, Zhejiang University, since September 2020. Before joining ZJU, he was a Chief Data Scientist in Verchable Ltd (UK). He got his PhD degree from Imperial College London and was supervised by Peter R.N. Childs (FREng). His research Interests: Design Intelligence, Computational creativity, UI Design, Human-computer interaction, and Artificial intelligence.

Authors:

Yuan Yin Imperial College London
Kaitong Qin Zhejiang University
Huiting Liu Zhejiang University
Lingyun Sun Zhejiang University
Peter Childs Imperial College London
Liuqing Chen Zhejiang University