Faces in advertising: exploring the interplay of attractiveness, realness categorisation, and purchase intentions

• Main Author: Koh, Jing Han
• Other Authors: Xu Hong
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Social Sciences; Advertisements; AI faces; Facial attractiveness; Purchase intentions; Marketing; AI
• Online Access: https://hdl.handle.net/10356/177872
• Institution: Nanyang Technological University

While digital ad spending is projected to soar in Singapore, a new player emerges – artificial intelligence. Powerful AI tools like Generative Adversarial Networks (GANs) promise significant cost savings in content creation. Particularly, the emergence of hyper realistic AI-generated faces may have the potential to upend the marketing industry. Past research has shown that faces are influential in advertising, with attractive faces increasing purchase intentions (PI), especially with paired with a related product. But can AI replicate this effect? To understand this, our study examined how FA, stimuli type (AI/Real faces), and participant categorisation (perceived AI/Real) affect PI. Employing a combination of Pixlr-generated AI faces and faces from an online research database, we presented these stimuli to 174 Singaporean participants (primarily Chinese undergraduates from NTU). PI was measured using a 4-item Purchase Intentions measure (PIMA). Our analyses (multiple regression and repeated-measures ANOVA) produced a surprising outcome: no significant difference in PI between AI and human faces in ads. Interestingly, even awareness of the manipulation (AI/Real) did not alter this effect. No interaction effects were found. However, FA remained significant: more attractive faces – whether AI or real – increased PI. Demographic factors, such as race and gender, subtly influenced PI, adding additional complexity to ad outcomes. Through this exploration, we hope to inform marketers on the suitability of cost-effective generative strategies as advertising alternatives. Future research may explore demographic influences on PI towards AI-aided ads and work towards validating these findings in real-world applications, such as through A/B testing.