Who Should Provide Clothing Recommendation Services: Artificial Intelligence or Human Experts?

Who Should Provide Clothing Recommendation Services: Artificial Intelligence or Human Experts?

Ziyang Li (Department of Industrial Engineering, Tsinghua University, Beijing, China), Pei-Luen Patrick Rau (Department of Industrial Engineering, Tsinghua University, Beijing, China) and Dinglong Huang (Shenzhen Malong Artificial Intelligence Research Center, Shenzhen, China)
Copyright: © 2020 |Pages: 13
DOI: 10.4018/JITR.2020070107
OnDemand PDF Download:
Available
$37.50
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

In this study, we investigated users' subjective attitudes (e.g., acceptance and perceived ease of use) towards artificial intelligence (AI) recommendation services and compared them with human expert recommendation services before and after use. Our experiments used fashion street snaps and online product images. The obtained experimental results indicated that the acceptance of the human expert recommendation service was higher than that of the AI recommendation service before use; however, they were similar after use in the case of both fashion street snaps and online product images. Furthermore, the perceived trustworthiness and perceived expertise of the AI and human expert recommendation services were also the same. In terms of ease of use, the AI recommendation service was perceived to be easier to use than the human expert recommendation service.
Article Preview
Top

Introduction

Rapid developments in machine learning and computer vision technologies have enabled the use of artificial intelligence (AI) for visual clothing recommendation services (e.g., Thomassey & Zeng, 2018; Wong, 2018). Prior to these AI-based recommendation services, most professional clothing recommendations were typically provided by human fashion experts. Besides, the knowledge of AI recommendation services are collected and learnt from experts’ experience and knowledge of clothing (Yang et al., 2018). Even though AI can work in a manner like an expert does, users’ acceptance and subjective attitudes (e.g., perceived usefulness, perceived ease of use, and usage intention) toward these two recommendation services (AI and human experts) might differ.

Users’ subjective attitudes towards recommendation sources is related to the perceived characteristics of the sources. Previous studies found a recommendation source perceived to have both trustworthiness and expertise would generate the greatest acceptance and decision changes (Pornpitakpan, 2004; Stiff & Mongeau, 2016). However, it is unknown that whether users perceived AI with similar expertise and trustworthiness as human experts. Moreover, actual use of a recommendation service by users will also influence their subjective attitudes. Actual use of such services can lead to more real and meaningful responses by users. To date, little research has been conducted to compare investigate similarities and differences between AI and human expert recommendation services. Based on the information above, this study tried to explore (1) users’ acceptance and (2) perceived trustworthiness and expertise of AI and human expert recommendation service; (3) effects of actual use on (1) and (2); (4) differences of users’ subjective attitudes to AI and human expert recommendation service after use.

This paper provides answers to these questions by two experiments, in which users input their personal clothing images and received recommendations in the forms of fashion street snaps (experiment one: fashion street snap experiment) and online product images (experiment two: online product recommendation experiment). Fashion street snaps are social media images of fashionable citizens and celebrities, whereas online product images are images from online stores. This study recommended two types of images in two experiments through social media to help users show their subjective attitudes based on actual use. This paper is organized as follows. Next, we would introduce related work and propose research questions. Then, the methods and results of the two experiments are presented. At last, we would discuss and conclude this paper.

Complete Article List

Search this Journal:
Reset
Open Access Articles
Volume 15: 6 Issues (2022): Forthcoming, Available for Pre-Order
Volume 14: 4 Issues (2021)
Volume 13: 4 Issues (2020)
Volume 12: 4 Issues (2019)
Volume 11: 4 Issues (2018)
Volume 10: 4 Issues (2017)
Volume 9: 4 Issues (2016)
Volume 8: 4 Issues (2015)
Volume 7: 4 Issues (2014)
Volume 6: 4 Issues (2013)
Volume 5: 4 Issues (2012)
Volume 4: 4 Issues (2011)
Volume 3: 4 Issues (2010)
Volume 2: 4 Issues (2009)
Volume 1: 4 Issues (2008)
View Complete Journal Contents Listing