See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/374088276

I Am ChatGPT, the ultimate AI Chatbot! Investigating the determinants of

users' loyalty and ethical usage concerns of ChatGPT

Article in Journal of Retailing and Consumer Services · January 2024

DOI: 10.1016/j.jretconser.2023.103562

CITATIONS READS

2 41

2 authors, including:

Florent Nkoulou
Shenzhen University
6 PUBLICATIONS 27 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Brand evangelism View project

All content following this page was uploaded by Florent Nkoulou on 22 September 2023.

The user has requested enhancement of the downloaded file.

 Journal of Retailing and Consumer Services 76 (2024) 103562

Contents lists available at ScienceDirect

Journal of Retailing and Consumer Services

journal homepage: www.elsevier.com/locate/jretconser

I Am ChatGPT, the ultimate AI Chatbot! Investigating the determinants of

users’ loyalty and ethical usage concerns of ChatGPT
Ben Niu, Gustave Florentin Nkoulou Mvondo *
College of Management, Shenzhen University, Shenzhen, 518060, China

A R T I C L E I N F O A B S T R A C T

Keywords: This research aims to explore the determinants of users’ satisfaction and loyalty towards ChatGPT while also
Information quality investigating ethical concerns related to the usage of the artificial intelligence (AI) chatbot. For this purpose, the
Perceived coolness study develops a framework based on five models and theories (information system success, technology
Technology affinity
acceptance model, affinity theory, coolness theory, and posthumanism) as well as other important constructs
Posthuman ability
Loyalty
(user ethical perceptions and user ethical beliefs). Analysis of data collected from 456 actual ChatGPT users in
Ethical usage concerns the US reveals several key findings. First, information quality significantly and positively affects users’ satis­
faction, perceived usefulness, and coolness. Second, perceived usefulness, coolness, technology affinity, and
posthuman ability also have a positive impact on users’ satisfaction, which subsequently influences their loyalty
to the AI chatbot. Furthermore, the findings demonstrate that user ethical perceptions and beliefs negatively
moderate the relationship between satisfaction and loyalty. The main implication of this research is that brand
managers and programmers should regularly assess the chatbot’s performance to ensure that the information
provided is relevant, reliable, concise, and delivered promptly. This is because users highly value the quality of
information delivered by the AI chatbot. Additionally, they should prioritize the ethical aspect, as it directly
influences users’ satisfaction and loyalty towards the chatbot services.

1. Introduction been trained using an extensive volume of varied textual information

gathered by crawling the internet, which makes it possible to interact in
“ChatGPT is scary good. We are not far from dangerously strong AI." a conversational way, answer follow-up questions, admit mistakes,
(Elon Musk). challenge incorrect premises, and reject inappropriate requests (OpenAI,
2022). ChatGPT possesses the ability to revolutionize a number of in­
Recently, there has been a great deal of hype and excitement around
dustries, such as customer service, education, healthcare, and finance.
a newly developed technology known as ChatGPT (Dwivedi et al.,
By providing assistance and facilitating the automation of tasks that
2023). The artificial intelligence (AI) chatbot and writing assistant is
were traditionally performed by humans, the AI chatbot can enhance
significantly altering and revolutionizing the way people communicate
efficiency, boost productivity, lower expenses, and create fresh avenues
and interact with technology (Mvondo et al., 2023a). Thanks to its
for growth (McKinsey, 2022).
advanced language processing abilities, it can understand and respond
ChatGPT has been a sensational hit since its launch in November
to natural language in real-time. It can hold meaningful conversations
2022. It attracted 100 million users just two months after launching and
that feel natural and human-like and assist users with quick, accurate,
averaged 25 million daily visitors in January 2023, making it the fastest-
and entertaining responses that set it apart from other chatbots in the
growing consumer application ever (Similarweb, 2023). By comparison,
market (Science Focus, 2023).
it took the popular app TikTok about nine months after its global launch
ChatGPT is an AI-based, pre-trained language model developed by
to reach 100 million users and Instagram more than two years (Milmo,
OpenAI (Dwivedi et al., 2023), an American AI research laboratory. The
2023). Despite ChatGPT’s increasing popularity and disruptive nature,
AI chatbot is based on GPT3.5 architecture (i.e., third-generation
the literature lacks studies that have systematically examined the factors
generative pre-trained transformer), a deep neural network designed
influencing consumers to use the AI chatbot. Owing to this, empirical
for processing sequential data, such as text (Kooli, 2023). The model has

* Corresponding author.
E-mail addresses: [email protected] (B. Niu), [email protected] (G.F.N. Mvondo).

https://doi.org/10.1016/j.jretconser.2023.103562
Received 19 May 2023; Received in revised form 1 September 2023; Accepted 2 September 2023
0969-6989/© 2023 Elsevier Ltd. All rights reserved.
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

examination of the determinants of users’ satisfaction and loyalty to the AI service agents (Ashfaq et al., 2020; Wang et al., 2023), intelligent
AI chatbot becomes relevant. For this purpose, we rely on a sample of assistants (Rajaobelina et al., 2021), virtual agents (Araujo, 2018), and
456 actual ChatGPT users in the US and use structural equation e-service agents (Selamat and Windasari, 2021). Chatbots are commonly
modeling (SEM). used as company representatives to improve customer service, engage
This research considers information quality (IQ) as the major factor and entertain customers, and support businesses by either replacing or
affecting user satisfaction and loyalty. IQ was initially proposed in the augmenting human support agents with AI (Chung et al., 2020; Jin and
information system success (ISS) model developed by DeLone and Youn, 2021). They support users for various purposes, including infor­
McLean (1992). It has been extensively studied in various contexts, such mation seeking, handling search queries, and building social relation­
as in research on ubiquitous computing applications (Kim et al., 2009), ships (Ashfaq et al., 2020). Chatbots can replicate spoken (voice-based
location-based apps (Chen and Tsai, 2019; Wang and Lin, 2017), AI chatbots) and written (text-based AI chatbots) human dialogue,
e-learning system success (Al-Fraihat et al., 2020), and chatbot customer allowing users to communicate with AI agents as if they were engaging
services (Kwangsawad and Jattamart, 2022). According to Setia et al. with live agents (Gkinko and Elbanna, 2022; Konya-Baumbach et al.,
(2013), IQ is “the accuracy, format, completeness, and currency of in­ 2023; Zhou et al., 2023). ChatGPT and traditional chatbots differ in the
formation produced by digital technologies” (p. 268). IQ measures how training data and approach. ChatGPT is trained on a massive amount of
well information serves its intended purpose and meets the user’s ex­ publicly available information, enabling it to provide answers and in­
pectations. The higher the quality of information, the more trustworthy sights on various topics and inquiries, even if it does not have a valid
and valuable it is for making decisions, solving problems, and achieving answer (Dai et al., 2023). Furthermore, ChatGPT can assist writers by
objectives (Kim et al., 2009; Kwangsawad and Jattamart, 2022). generating human-like texts and even programming computers. Unlike
In addition to IQ, this research also takes into account other relevant other chatbots, which are trained on specific topics and simply provide
factors from established models and theories. These factors include pre-written responses, ChatGPT generates answers rather than just
perceived usefulness from the technology acceptance model (TAM), copying and pasting them (Loeppky, 2023).
technology affinity from the media affinity theory, perceived coolness In recent years, scholars have shown increased interest in examining
from the coolness theory, and posthuman ability from posthumanism. the relationship between humans and virtual agents. A review of the
Perceived usefulness, as defined by Davis (1989), refers to the extent to literature suggests that most studies have focused on comparing human-
which individuals believe that using a particular system would enhance chatbot and human-human communication (e.g., Hill et al., 2015; Mou
their job performance. It is a crucial factor in determining user accep­ and Xu, 2017), with most research employing the “computer are social
tance of technology. Technology affinity represents users’ perceived actors” paradigm, which was advanced by Nass et al. (1994). Addi­
importance of technology in their lives, reflecting their affinity towards tionally, Chatbots have been studied in various contexts, including
technology and its role in various aspects of their daily routines and healthcare services (Ben-Shabat et al., 2022), customer services (Tran
activities (Aldás-Manzano et al., 2009; Perse, 1986). Perceived coolness et al., 2021), finance (Huang and Lee, 2022), education (Celik, 2023),
encompasses the overall judgment of a technological product, including and travel services (Nguyen et al., 2022).
its perceived originality, attractiveness, and the potential to build a Scholars have also scrutinized the association between virtual agents
subculture around it (Sundar et al., 2014). Coolness plays a significant and users’ satisfaction. For example, Ashfaq et al. (2020) studied the
role in shaping user perceptions and attitudes toward technology determinants of user satisfaction with chatbots, taking to account the ISS
(Mamonov and Koufaris, 2020; Sundar et al., 2014). Posthuman ability model, expectation-confirmation model, and TAM. They found that
refers to the enhanced capabilities or attributes that surpass those of these factors positively affect user satisfaction. Hsu and Lin (2023)
typical human beings (Nath and Manna, 2023). It represents the ability analyzed consumer satisfaction and loyalty toward e-service agents and
of technology to perform a specific task better than a human (Gambino found that AI chatbot service recovery quality and AI chatbot conver­
and Sundar, 2019). sational quality significantly influence user satisfaction. Moreover,
Earlier studies have argued that ethical concerns may have an impact Nguyen et al. (2022) compared user interactions with chatbot interfaces
on users’ attitudes toward technology (Cheung and To, 2021; Nadeem and menu-based interfaces; they found that chatbot systems lead to a
et al., 2020; Román, 2007). ChatGPT is a versatile technology with a lower level of perceived autonomy and higher cognitive load compared
wide range of applications. However, concerns have been raised with menu-based interface systems, resulting in a lower degree of user
regarding unethical practices associated with the AI chatbot. Therefore, satisfaction.
this research also examines the interaction effect of user ethical per­ Despite the growing number of studies on chatbots, the determinants
ceptions of ChatGPT and user ethical beliefs on the relationship between of user satisfaction and loyalty are still limited. Additionally, the existing
satisfaction and loyalty. literature lacks studies that provide a comprehensive understanding of
In summary, the objective of this research is three-fold: (1) To the relationship between IQ, satisfaction, and loyalty. Earlier studies
develop a novel integrated framework that provides a comprehensive have investigated the association between IQ and satisfaction, but they
understanding of the determinants of user satisfaction and loyalty to­ adopted a holistic perspective and did not shed light on the underlying
wards ChatGPT. (2) To conduct an empirical analysis to validate the mechanisms (i.e., factors) of IQ and how they impact user satisfaction (e.
conceptual model. (3) To understand the potential moderating role of g., Ruan and Mezei, 2022; Tisland et al., 2022). IQ is a complex and
user ethical perceptions of ChatGPT and user ethical beliefs on the multidimensional phenomenon (Ge et al., 2011); accordingly, it is crit­
relationship between satisfaction and loyalty. The findings of this ical to determine the key aspects important to customers to help com­
research will offer valuable guidance to programmers and brand man­ panies develop effective strategies for improving the quality of
agers involved in the development and management of AI chatbots. By information.
understanding the factors that contribute to user satisfaction and loy­ To extend the literature on chatbots and fill the gap, the current
alty, as well as the potential influence of ethical concerns, programmers study proposes a model based on the ISS model by including IQ and its
and brand managers can make informed decisions to enhance user ex­ underlying factors comprising relevance, reliability, conciseness, and
periences and foster stronger relationships with their users. quickness. The model also incorporates other factors from models and
theories such as TAM (perceived usefulness), media affinity theory
2. Literature review on chatbots (technology affinity), coolness theory (perceived coolness), and post­
human ability (posthumanism). Additionally, the model includes two
Shawar and Atwell (2005) defined chatbots as “machine conversa­ new constructs: user ethical perceptions of ChatGPT and user ethical
tion systems [that] interact with human users via natural conversational beliefs, which serve as moderators in the relationship between satis­
language” (p. 489). They are identified under different names, including faction and loyalty. Users’ ethical perceptions and beliefs can influence

2
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

their behavior (Cheung and To, 2021; Nadeem et al., 2020). Ajzen and 3.2. Technology acceptance model: perceived usefulness
Fishbein (2005) conducted a comprehensive analysis of
attitude-behavior research, wherein they observed that while overall TAM is a popular model in IS research. The model was proposed by
attitudes can significantly impact behavior, this influence is contingent Davis (1985, 1989) to explain how consumers come to adopt computer
upon specific circumstances or individual characteristics. In other technology and has been widely used in various contexts, such as in
words, the relationship between attitude (i.e., satisfaction) and behavior studies on wearable healthcare technology (Kim and Ho, 2021), video
(i.e., loyalty) is generally moderated by factors pertaining to the indi­ game consoles (Nan et al., 2022a), virtual reality (Seong and Hong,
vidual performing the behavior. Ethical perceptions and beliefs are 2022), and fashion mobile apps (Pop et al., 2023). TAM suggests that the
considered to be plausible factors that reflect individual differences acceptance of an IS by users depends on two key factors: perceived
(Cheung and To, 2021; Román, 2007). Hendriana et al. (2013) argued usefulness and perceived ease of use. The use of an IS is determined by “a
that ethical beliefs represent opinions that users adhere to in order to behavioral intention to use the system, where the intention to use IS is
form a generally accepted standard of conduct, and Román (2007) determined by an individual’s attitude toward using the system and its
suggested that ethical perceptions refer to how users perceive a com­ perceived usefulness” (Verma et al., 2018, p. 3). This research only
pany’s responsibility, integrity, and commitment to ensuring secure, adopted the concept of perceived usefulness because it is believed that
honest, fair, and confidential treatment of users, with a priority on ChatGPT is user-friendly, as users only need to type a question or paste
consumer protection. To the best of the authors’ knowledge, previous content. This research defines perceived usefulness as an individual’s
research has not examined the interaction effect of user ethical per­ perception of how AI chatbots can provide practical benefits, improve
ceptions and user ethical beliefs on the satisfaction-loyalty relationship. efficiency, increase productivity, or enhance overall effectiveness. It is
The inclusion of these two constructs will enhance the model’s reli­ the perception that an IS is beneficial in helping an individual achieve
ability and provide a deeper understanding of ethical concerns sur­ their work-related goals.
rounding ChatGPT.
3.3. Media affinity theory: technology affinity
3. Theoretical foundation
The media affinity theory is defined as “the importance of the me­
3.1. Information system success: information quality dium in the lives of individuals” (Aldás-Manzano et al., 2009, p 741).
This theory is grounded in uses and gratifications research, which ex­
In today’s fast-paced information-driven world, businesses and in­ plains that individuals actively seek and consume media content with
dividuals rely heavily on information to make informed decisions and the objective of gratifying their needs and satisfying their various in­
take effective actions. IQ measures the quality of the content of an IS terests (Perse, 1986; Rubin, 1981). This research conceptualizes tech­
(Ma, 2021; Zhong and Chen, 2023; Zrnec et al., 2022). This study defines nology affinity as users’ perceived importance of AI chatbots in their life.
IQ as the quality of the information provided by AI chatbots. IQ is critical Media affinity has been used to evaluate individuals’ attitudes toward
because it directly affects decision-making accuracy and effectiveness technology. For instance, Xu and Du (2018) studied the determinants of
(Cheng et al., 2021; Demoulin and Coussement, 2020; Kumar, 2023). user loyalty to digital libraries and found that IQ, service quality, and
High-quality information is vital to users and helps them save time and system quality are critical determinants of users’ affinity to digital li­
effort. Poor IQ may result in incorrect decisions and lower competi­ braries. Likewise, Aldas-Manzano et al. (2009) explored the drivers of
tiveness (Hausvik et al., 2021; Song et al., 2022; Torres and Sidorova, mobile shopping acceptance and found that mobile affinity increases
2019). usage intentions.
IQ is a multidimensional construct that encompasses important re­
lationships among multiple factors. In a study on data quality manage­ 3.4. Coolness theory: perceived coolness
ment, Wang and Strong (1996) classified IQ into four groups: contextual,
intrinsic, accessibility, and representational. Meanwhile, in a study on The coolness theory, also known as perceived coolness, was proposed
ubiquitous computing applications, Kim et al. (2009) identified three by Sundar et al. (2014) to explain the perception of coolness in tech­
dimensions of IQ: relevance, accuracy, and timeliness. Ge et al. (2011) nological products. They defined perceived coolness as an overall
further assessed IQ by distinguishing between raw data stored in data­ judgment of a technological product that is perceived as original,
bases and the information product delivered to users and identified nine attractive, and capable of building a subculture around it (Sundar et al.,
dimensions of IQ. This research builds upon earlier studies by Wang and 2014). A product’s coolness contributes to its success in the marketplace
Strong (1996), Kim et al. (2009), and Ge et al. (2011) and identifies four because it plays a critical role in how consumers view and use it.
dimensions of IQ that are relevant to the study context: relevance, Coolness is subject-specific; consumers may perceive a product’s cool­
reliability, conciseness, and quickness. Relevance of information refers to ness differently (Warren et al., 2018). Several scholars assert that when
“the correspondence between users’ requests and system responses or consumers perceive a product/service as cool, they are more likely to
between users’ expectations and service” (Kim et al., 2009, p. 440). show positive behavioral intentions, such as loyalty (Chen and Chou,
Reliability of information refers to the trustworthiness and dependability 2019), adoption intention (Mamonov and Koufaris, 2020), usage
of information (Ge et al., 2011). Conciseness of information is the degree intention (Ahn and Park, 2022), and purchase intentions (Reinikainen
to which the information is well-formatted, well-organized, and et al., 2021). Perceived coolness is a complex and multidimensional
aesthetically appealing (Wang and Strong, 1996). Information provided phenomenon “comprising attractiveness, subcultural appeal, and origi­
by an information system (IS) should be clear, concise, and nality.” Attractiveness denotes the aesthetic of a product. Sundar et al.
well-presented (Ge et al., 2011). In this research, we modified the term (2014) pointed out that people tend to be attracted to products with
“timeliness” to “quickness.” This is necessary because ChatGPT does not aesthetically appealing features and functions. A subcultural appeal de­
provide up-to-date information, as it has limited knowledge of the world notes products whose usage promotes the formation of subcultures.
and events beyond 2021 (OpenAI, 2023). As a result, the AI chatbot According to Nan et al. (2022a), “since the use of products that are
cannot answer current questions. We then define quickness of information unique in appearance and function is uncommon in mainstream society,
as the speed at which users obtain information from ChatGPT. Quick their use tends to stand out, promoting the development of subcultures
access to information is highly valued, as it can significantly impact related to these products” (p. 3). Originality is related to a product’s
decision-making, productivity, and overall efficiency. uniqueness and innovativeness. Products with unique designs and
functions are generally perceived as original (Ahn and Park, 2022;
Sundar et al., 2014).

3
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

3.5. Posthumanism: posthuman ability conciseness, and d) quickness is related to perceived coolness.
User satisfaction is “an established common measure of the success
Posthumanism is one of the well-known theories in the present day
or effectiveness of IS in general” (Ghasemaghaei and Hassanein, 2015, p.
(Nath and Manna, 2023). In the book titled “Why I want to be a post­
7). The ISS model posits that “the quality of information as an output of
human when I grow up?” Bostrom (2013) argues that being “posthu­
an IS is one of the major components explaining user satisfaction”
man” means attaining at least one posthuman ability. He further
(Laumer et al., 2017, p. 336). Individuals use an IS with the expectation
explains posthuman ability as a competence that far surpasses the
that it helps them accomplish a task, and overall satisfaction is the main
maximum achievable level for any human being without the assistance
factor determining whether they will transact with it and return to it in
of smart technology. In IS research, posthuman ability has been
the future (Arghashi and Yuksel, 2022; Ashfaq et al., 2020; Ghasema­
conceptualized as the capacity of technology to outperform humans in a
ghaei and Hassanein, 2015). In a study on users of enterprise content
specific task (Gambino and Sundar, 2019; Nath and Manna, 2023).
management systems, Laumer et al. (2017) revealed that when infor­
Intelligent systems capable of surpassing human intelligence in a
mation is not relevant to a task or is difficult to understand, individuals
particular field are considered posthuman (Nath and Manna, 2023).
may seek alternative ways to obtain the information. In contrast, when
The introduction of smart technologies with innovative features
an IS provides quick, relevant, reliable, and easily understandable in­
often leads to tensions as they take up roles traditionally reserved for
formation, it can enhance users’ satisfaction. Based on this reasoning,
humans. This is often seen through a pessimistic lens. However, “how
we propose that:
well a technology can perform its job may combat negative factors”
(Gambino and Sundar, 2019, p. 2). For instance, calculators, which are H3. Information quality dimensions of a) relevancy, b) reliability, c)
simple devices designed to perform mathematical calculations previ­ conciseness, and d) quickness is related to user satisfaction.
ously done mentally or on paper, are defined by their ability to
considerably outperform humans. Autonomous vehicles are another 4.2. Perceived usefulness and user satisfaction
example of technology perceived as posthuman (Gambino and Sundar,
2019). Although individuals may have issues and very important ethical Perceived usefulness and satisfaction are two major factors that
dilemmas regarding driverless cars, they may still conclude that the determine the success of any IS, including ChatGPT. The perceived
technology is acceptable based on the sheer abilities of the driverless car usefulness of a chatbot can be measured by its ability to provide relevant
and its network (Gambino and Sundar, 2019). and helpful information, solve problems, and assist users in achieving
their objectives (Van den Broeck et al., 2019; Kasilingam, 2020; Selamat
4. Hypotheses development and theoretical framework and Windasari, 2021). Earlier research has examined the link between
usefulness and satisfaction. For instance, Joo et al. (2017) found that
4.1. Information quality, perceived usefulness, perceived coolness, and digital textbooks’ perceived usefulness is related to user satisfaction.
user satisfaction Similarly, Li and Fang (2019) found that usefulness is a critical predictor
of users’ satisfaction. Thus, we propose that:
Individuals invest significant time and effort seeking out answers to
H4. Perceived usefulness has a positive effect on user satisfaction.
specific questions, advice on various topics, explanations of complex
concepts, recommendations for products or services, opinions on current
4.3. Technology affinity and user satisfaction
events, and assistance with problem-solving (Ashfaq et al., 2020). Thus,
virtual agents should provide personalized, complete,
As explained earlier, technology affinity is users’ perceived impor­
easy-to-understand, and well-formatted information because IQ is one of
tance of technology in their life. Several scholars have suggested that the
the most important quality components to measure the success of a
stronger a user’s relationship with technology, the higher the likelihood
system (Teo et al., 2008).
they feel that using it meets their expectations (Aldás-Manzano et al.,
Since its launch, ChatGPT has become a sensational hit because it can
2009). This is because a positive affinity towards technology fosters
generate human-like responses to a wide range of queries and topics. It
greater trust and comfort, resulting in a more positive overall experience
can understand natural language inputs from users and provide quick,
(Perse, 1986). In the context of ChatGPT, if users have a positive affinity
relevant, and reliable (in most cases) information that is easy to un­
towards the AI chatbot, they are more likely to be satisfied with their
derstand. IQ can impact the perceived usefulness of an IS. Demoulin and
interactions. In contrast, users with a negative affinity toward ChatGPT
Coussement (2020) argued that high-quality information enhances
are more likely to be dissatisfied. Negative experiences can lead to
users’ and businesses’ performance and efficiency. The factors of rele­
frustration and mistrust, which can further decrease the user’s affinity
vance, reliability, conciseness, and quickness are important for deter­
for the AI chatbot.
mining the quality of information. Information that is well-organized,
Earlier studies have found a strong relationship between users’ af­
trustworthy, accurate, match users’ needs or task, and is quickly
finity to technology and satisfaction. Soltani-Nejad et al. (2020) found
accessible is more likely to be perceived as useful. Earlier research has
that users who have a closer relationship with technology are more
found that IQ positively influences a digital device’s perceived useful­
likely to develop an affinity for it and, as a result, tend to be more
ness (Xu and Du, 2018). Accordingly, we propose that:
satisfied with the technology. Likewise, Xu and Du (2018) found that
H1. Information quality dimensions of a) relevancy, b) reliability, c) users’ affinity for digital libraries increases satisfaction. To echo earlier
conciseness, and d) quickness is related to perceived usefulness. research findings, we hypothesize that:
The perception of coolness is subject-specific (Warren et al., 2018). H5. Technology affinity positively influences user satisfaction.
The association between IQ and coolness is not straightforward, as it
depends on users’ preferences. Generally, if the information provided is 4.4. Perceived coolness and user satisfaction
reliable and corresponds to users’ needs or interests, it is more likely to
be perceived as cool (Peng et al., 2016). ChatGPT can, in most cases, In recent years, perceived coolness has emerged as a critical
provide high-quality information on a wide range of queries and topics. construct in IS research. Coolness is the strongest positive expression
Therefore, we assume that users may perceive the information provided validating a technological device’s novelty, originality, and attractive­
by the AI chatbot as cool. ness (Chen and Chou, 2019; Nan et al., 2022a). A chatbot that is
perceived as cool is more likely to be engaging and enjoyable to interact
H2. Information quality dimensions of a) relevancy, b) reliability, c)
with, which can increase user satisfaction. Similarly, a chatbot that

4
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

effectively meets a user’s needs and provides a satisfying experience is protection to consumers’ interests” (p. 3). Privacy is the primary concern
more likely to be perceived as cool by the users. Earlier research has for users when interacting with technology (Nadeem et al., 2020). While
demonstrated that coolness is an important determinant of user satis­ users perceive ChatGPT as enjoyable and useful, they may also view it as
faction (Liu and Mattila, 2019; Nan et al., 2022b). Therefore, we propose potentially risky in terms of ethical lapses. When interacting with the AI
that: chatbot, users provide sensitive information that becomes part of the
public domain. Additionally, OpenAI collects a wide range of user in­
H6. Perceived coolness positively influences user satisfaction.
formation, including IP address, browser type and settings, and data on
users’ interactions with the site (Smith, 2023; Veale, 2023). These
4.5. Posthuman ability and user satisfaction concerns can influence users’ loyalty. Thus, we predict that user ethical
perceptions will moderate the association between satisfaction and
Digital devices that outperform humans in specific tasks are loyalty, resulting in decreased user loyalty to ChatGPT. Accordingly, we
perceived as posthuman (Gambino and Sundar, 2019). ChatGPT, an hypothesize that:
AI-powered chatbot, outperforms humans by providing accurate an­
swers to a broad range of queries and topics. In addition, the AI chatbot H9. The association between satisfaction and loyalty is negatively
can program computers, which has helped users who were struggling to moderated by user ethical perceptions.
write code. Gambino and Sundar (2019) argued that “the efficiency or OpenAI has not been transparent about the algorithm and data
the ability of a device to perform tasks at a very high level may be the source used to train ChatGPT (McGowan, 2023). The model is trained on
key to the public acceptance of the technology” (p. 2). We predict that data from the internet, which may contain biases, misinformation, or
ChatGPT’s ability to outperform humans in various tasks will lead to a controversial content. As a result, there are instances where the AI
positive user experience. Accordingly, we hypothesize that: chatbot inadvertently learns and reproduces these biases and inaccura­
H7. Posthuman ability has a positive effect on user satisfaction. cies in its responses (e.g., stereotypes, race, and gender discrimination)
(Germain, 2023).
4.6. User satisfaction and loyalty Users’ ethical beliefs can significantly influence their behavior
(Cheung and To, 2021). As mentioned earlier, user ethical beliefs refer to
Jones and Sasser (1995) defined loyalty as “a feeling of attachment the opinions that users adhere to in order to establish a generally
or affection for a company’s people, products, or services” (p. 94). Ac­ accepted standard of conduct (Hendriana et al., 2013). Individuals with
cording to Oliver (1999), loyalty is “a deeply held commitment to rebuy strong ethical beliefs tend to apply strict moral standards when evalu­
or repatronize a preferred product/service consistently in the future” (p. ating a company’s conduct or its products (Vitell and Paolillo, 2004).
34). These two definitions suggest that loyalty is a multidimensional These individuals are particularly attentive to whether a company pri­
construct comprising attitudinal and behavioral loyalty. In fact, several oritizes corporate social responsibility as an integral aspect of its oper­
researchers have argued that loyalty encompasses two dimensions: ations (Vitell and Paolillo, 2004). When users perceive that AI firms have
continuance intention and WOM (Chiu et al., 2013; Xu et al., 2015; embraced ethical practices such as transparency in the data used to train
Thakur, 2018). Other researchers have adopted a composite approach AI chatbots and accountability, they may perceive these practices as
by integrating both attitudinal and behavioral loyalty into a measure of aligning with their own stringent moral codes, which aim to protect
loyalty. For instance, Liu et al. (2011) integrated the items measuring others and society. Individuals identifying with a company’s ethical
recommendation and continuance usage into a measure of loyalty. This practices are likely to reflect those practices through positive behaviors
research adopts this perspective and integrates continuance usage and (Cheung and To, 2021), such as loyalty. Conversely, individuals who
WOM in the measure of loyalty. We conceptualize loyalty as a feeling of perceive a company’s practices as unethical will not endorse or support
attachment to the chatbot’s information resources and related services, those practices. Based on this reasoning. We predict that user ethical
characterized by continued usage and positive WOM. beliefs will moderate the association between satisfaction and loyalty,
Individuals who develop a bond with a brand will remain loyal and resulting in decreased user loyalty to ChatGPT.
engage in supportive behavior (Mvondo et al., 2023b). User satisfaction H10. The association between satisfaction and loyalty is negatively
with ChatGPT represents an accumulative feeling developed throughout moderated by user ethical beliefs (Fig. 1).
the process of using the AI chatbot and interacting with it. Several
scholars have asserted that satisfaction and loyalty are closely related 5. Methodology
(Nan et al., 2022b). In a study on mobile commerce, Sarkar et al. (2020)
asserted that when users are satisfied with technology, they are more 5.1. Measures
likely to remain loyal and continue using the service over time. Simi­
larly, in research on customer service chatbots, Hsu and Lin (2023) The research instrument and items were developed by referring to
found that user satisfaction positively impacts loyalty. To echo previous existing literature and slightly adjusted to fit the study context. A 7-point
research findings, we advance that: Likert-type scale was used for measuring the items, with a rating scale
H8. User satisfaction positively influences loyalty. ranging from “1 (strongly disagree) to 7 (strongly agree).” To assess IQ, a
set of eleven items was adapted from the studies conducted by Ge et al.
(2011), Kim et al. (2009), and Wang and Strong (1996). The scale
4.7. Ethical concerns of ChatGPT
consisted of three items for each relevance, reliability, and conciseness
dimension. We revised the “timeliness” dimension to “quickness,”
Ethical issues arise when a particular decision, scenario, or activity
removed the item “up to date,” as it was irrelevant to this study, and
conflicts with the moral principles of a society. Both individuals and
assessed it using a two-item scale.
businesses can be involved in these conflicts, as any of their actions may
Perceived usefulness was assessed using a three-item scale adapted
be subject to ethical scrutiny (Indeed, 2022). With respect to ChatGPT,
from Davis (1989). Technology affinity was measured using a three-item
two types of ethical concerns can arise from its use: user ethical per­
scale adapted from Xu and Du (2018). Perceived coolness was evaluated
ceptions of the AI chatbot and user ethical beliefs.
using a six-item scale adapted from Sundar et al. (2014). Posthuman
According to Román (2007), consumer ethical perceptions refer to
ability was measured using a single item adapted from Gambino and
"consumers’ perceptions of a company’s responsibility and integrity
Sundar (2019). Satisfaction was assessed using a three-item scale
behind the online platform in its attempt to deal with consumers in a
adapted from Oliver (1980), while loyalty was measured using a
secure, honest, fair, and confidential manner, ultimately giving

5
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

Fig. 1. Theoretical model.

four-item scale adapted from Zhao et al. (2012) and Bloemer et al.
Table 1
(1999). User ethical beliefs were evaluated using a five-item scale
Demographic characteristics.
adapted from Chaudhry and Stumpf (2011) and Cheung and To (2021),
and user ethical perceptions were measured using a five-item scale Items Frequency (N = 456) (%)

adapted from Román (2007). We conducted a small-scale pilot study on Gender

41 ChatGPT users, and based on their feedback, we further revised and Male 284 62.3
Female 172 37.7
finalized the questionnaire. The reliability scale of all the constructs
Age
exceeded 0.50, which was considered acceptable. 18–25 Years 138 30.3
26–35 Years 118 25.9
5.2. Sample collection 36–45 Years 97 21.3
46–55 Years 63 13.8
55 Years or above 40 8.8
The authors gathered data from US residents using a structured on­
Education
line questionnaire administered through Amazon Mechanical Turk High School or Less 48 10.5
(MTurk). MTurk is a widely-used platform among scholars due to its Bachelors 232 50.9
high-quality responses and the naivety of its respondents. Additionally, Masters 140 30.7
Doctor (PhD) 36 7.9
it enables researchers to collect data from diverse locations throughout
Others 80 9.2
the US. Profession
We administered the survey between February and March 2023. The Student 171 37.5
questionnaire was composed of three parts. First, we explained the Employee 146 32
purpose of the research. Second, we collected demographic information, Businessman/Businesswoman 101 22.1
Unemployed 30 6.6
including a question about respondents’ frequency of ChatGPT use. The
Others 8 1.8
final section included the variables of interest. We proceeded with the How often do you use ChatGPT
data collection phase using a purposive sampling technique and a cross- More than Once a day 197 43.2
sectional survey. We specifically targeted respondents who had used Once a day 73 16
More than once a week 58 12.7
ChatGPT by incorporating a screening question that asked if they had
Once a week 52 11.4
used ChatGPT within the past three months. Those who answered “No” More than once a month 33 7.2
were directed to the end of the survey. A total of 456 valid responses Once a month 26 5.7
were used for data analysis and comprised 284 men (62.3%) and 172 Just on special occasions 17 3.7
women (37.7%), 77.2% of the respondents were aged between 18 and
45, 89.5% held at least an undergraduate degree, and 60.7% were em­ 6. Results
ployees or businesspeople. Furthermore, 59.2% of the respondents re­
ported using ChatGPT at least once a day (see Table 1). 6.1. Common method bias

5.3. Data analysis To address CMB, the questionnaire included instructions on the strict
anonymity protocol of the study. Respondents were told to be neutral
The authors used IBM SPSS 27.0 to assess common method bias and honest and were informed that there were no right or wrong an­
(CMB) and the demographics and Amos 24.0 to assess the measurement swers. We also performed Harman’s single-factor approach, and the
and structural model. Data were analyzed in two phases: (1) model fit, total variance for a single factor was 14.01%, which met the cut-off
data reliability, and validity, and (2) hypotheses testing. criteria. Furthermore, we conducted a full collinearity test, following
Kock (2015) ‘s recommendation. The results demonstrated that all the
variance-inflated factors were lower than 3.3. Thus, CMB is not of
concern in this study.

6
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

6.2. Measurement validity Table 2

Reliability and validity.
We measured the model fitness, construct reliability, and convergent Construct Items Loading CR AVE
and discriminant validity. The model delivered adequate fit indices as
the value of chi-square (x2/df = 1.053), incremental fit index (IFI =
>0.7 >0.5

0.997), comparative fit index (CFI = 0.997), Tucker Lewis index (TLI = Relevance of RVI1: The information provided 0.870*** 0.88 0.711
Information by ChatGPT is useful.
0.997), normed fit index (NFI = 0.947), standardized root mean square RVI2: The information provided 0.822***
residual (SRMR = 0.0266), and root mean square error of approximation by ChatGPT is relevant.
(RMSEA = 0.011) were satisfactory. RVI3: The information provided 0.836***
Moreover, the constructs’ composite reliability (CR) value exceeded by ChatGPT is applicable.
Reliability of RI1: The information provided 0.878*** 0.893 0.736
0.70, establishing the measurement instruments’ reliability. The outer
Information by ChatGPT is reliable.
loadings were above 0.704, which proved to be acceptable. The average RI2: The information provided 0.845***
variance extracted (AVE) was confirmed as all items had factor loadings by ChatGPT is trustworthy.
above the threshold of 0.50. Additionally, the Fornell-Larcker criterion RI3: The information provided 0.850***
and the Heterotrait-Monotrait (HTMT) ratio showed that discriminant by ChatGPT is credible.
Conciseness of CI1: The information provided 0.863*** 0.899 0.748
validity was established in the study (see Tables 2 and 3). Information by ChatGPT is well-presented.
CI2: The information provided 0.867***
6.3. Structural model by ChatGPT is aesthetically
pleasing.
CI3: The information provided 0.865***
The proposed model and hypothesized relationships were analyzed
by ChatGPT is well formatted.
using the R-squared method and evaluating the level and significance of Quickness of QI1: ChatGPT provides prompt 0.871*** 0.885 0.794
the path coefficients. R-squared is the proportion of an endogenous information information.
construct’s variance explained by its predictor constructs in a regression QI2: ChatGPT provides 0.911***
model (Mvondo et al., 2022a, 2022b). The results showed that the interesting information without
any delay.
endogenous constructs, namely, perceived usefulness (0.564), perceived Perceived PU1: Overall, ChatGPT is useful 0.862*** 0.89 0.729
coolness (0.526), satisfaction (0.515), and loyalty (0.628), had large usefulness for me.
effects, indicating that the regression model was acceptable. PU2: ChatGPT improves the 0.852***
We found that the IQ factors of relevance (β = 0.291, 0.300, and quality of my daily life.
PU3: ChatGPT is a useful device 0.847***
0.104, i.e., perceived usefulness, coolness, and satisfaction), reliability
in my life.
(β = 0.299, 0.295, and 0.093), conciseness (β = 0.186, 0.128, and Perceived PC1: ChatGPT is cool. 0.878*** 0.941 0.725
0.097), and quickness (β = 0.135, 0.156, and 0.126) positively influence coolness PC2: When I first heard of 0.843***
perceived usefulness, perceived coolness, and satisfaction. Additionally, ChatGPT, I remember thinking it
we found that perceived usefulness (β = 0.136), technology affinity (β = would be cool to use it.
PC3: Using ChatGPT make me 0.832***
0.169), perceived coolness (β = 0.093), and posthuman ability positively look cool.
influence user satisfaction, which, in turn, increases user loyalty (β = PC4: When I use ChatGPT, my 0.854***
0.597) (see Table 4). The findings also revealed that user ethical per­ response is often like, “That’s
ceptions (β = − 0.216) and User ethical beliefs (β = − 0.168) have a cool!”
PC5: ChatGPT has some cool 0.849***
negative moderating effect on the association between satisfaction and
features.
loyalty (see Tables 4 and 5, Figs. 2 and 3). PC6: ChatGPT is cooler than 0.853***
other chatbots.
7. Discussion Satisfaction SAT1: I am satisfied with my 0.862*** 0.903 0.756
decision to use ChatGPT.
SAT2 I am satisfied with my 0.872***
This research examined the key determinants of user satisfaction and
previous experiences with
loyalty towards ChatGPT while also investigating ethical concerns ChatGPT.
related to the usage of the AI chatbot. To achieve this, an integrated SAT3: My choice to use ChatGPT 0.874***
model based on five models and theories (ISS, TAM, affinity theory, is a wise one.
Technology TA1: Seeking information on 0.864*** 0.881 0.712
coolness theory, and posthumanism) was proposed, along with other
affinity ChatGPT is one of my main daily
important constructs such as user ethical perceptions and user ethical activities.
beliefs. The findings confirmed that the IQ dimensions of relevance, TA2: ChatGPT is important in 0.836***
reliability, conciseness, and quickness have a significant and positive my life.
impact on perceived usefulness, perceived coolness, and satisfaction TA3: I cannot go for several days 0.831***
without using ChatGPT.
(H1, H2, H3). Additionally, perceived usefulness, technology affinity,
User loyalty UL1: I intend to continue using 0.836*** 0.908 0.713
perceived coolness, and posthuman ability were found to positively in­ ChatGPT in the future and will
fluence user satisfaction (H4, H5, H6, H7), which, in turn, increased user keep using it regularly as I do
loyalty (H8). As predicted, user ethical perceptions and beliefs nega­ now.
UL2: I intend to increase my use 0.863***
tively moderated the association between satisfaction and loyalty (H9
of ChatGPT in the future.
and H10). These findings suggest that when the AI chatbot provides UL3: I will say positive things 0.843***
high-quality information and is perceived as useful and cool, users are about ChatGPT to other people.
more satisfied and likely to remain loyal. However, user ethical per­ UL4: I will strongly recommend 0.835***
ceptions and beliefs can negatively affect the relationship between others to use ChatGPT
User ethical UEP1: I think ChatGPT collects 0.794*** 0.89 0.617
satisfaction and loyalty, highlighting the need to address ethical con­
perceptions my personal information.
cerns to maintain user loyalty. UEP2: I recognize that disclosing 0.764***
personal information through
ChatGPT is a risk.
(continued on next page)

7
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

Table 2 (continued )
Construct Items Loading CR AVE
Table 4
>0.7 >0.5
Direct effects.
UEP3: ChatGPT uses collected 0.814***
information for research Hypothesized relationships Std. Std.
purposes Beta Error
UEP4: I think that ChatGPT may 0.774***
disclose my information to third H1a Relevance→Perceived usefulness 0.291** 0.049
parties without consent. H1b Reliability→Perceived usefulness 0.299** 0.042
UEP5: ChatGPT denies users the 0.781*** H1c Conciseness→Perceived usefulness 0.186*** 0.042
right to choose if they wish their H1d Quickness→Perceived usefulness 0.135*** 0.033
data to be collected. H2a Relevance→Perceived coolness 0.300*** 0.043
User ethical UEB1: A person should ensure 0.774*** 0.885 0.607 H2b Reliability→Perceived coolness 0.295** 0.038
beliefs that his/her actions never H2c Conciseness→Perceived coolness 0.128** 0.037
intentionally harm others. H2d Quickness→Perceived coolness 0.156*** 0.030
UEB2: A person should never 0.789*** H3a Relevance→Satisfaction 0.104* 0.041
psychologically harm others. H3b Reliability→ Satisfaction 0.093* 0.042
UEB3: A person should not 0.785*** H3c Conciseness→ Satisfaction 0.097* 0.045
perform an action that might in H3d Quickness→ Satisfaction 0.126* 0.050
any way threaten the dignity and H4 Perceived usefulness→Satisfaction 0.136** 0.051
welfare of others. H5 Technology affinity →Satisfaction 0.169** 0.060
UEB4: A person should not harm 0.756*** H6 Perceived coolness→Satisfaction 0.093* 0.045
innocent others. H7 Posthuman ability→Satisfaction 0.106* 0.047
UEB5: A person should not 0.791*** H8 Satisfaction→Loyalty 0.597** 0.027
sacrifice the welfare of others.

Table 5
8. Implications and future research lines
Interaction effect.
Hypothesized relationships Std. Std.
8.1. Theoretical implications
Beta Error
This study has several academic implications. First, most research on H9 InterSATxUEP − 0.216*** 0.029
ChatGPT has primarily focused on its application in education. For H10 InterSATxUE − 0.168** 0.028
instance, Tlili et al. (2023) explored the use of ChatGPT in education
among early adopters, while Arif et al. (2023) and Eysenbach (2023)
examined its application in medical education. This paper is unique in
that it pioneers the identification of the determinants of users’ loyalty
and ethical usage concerns of ChatGPT. In other words, we integrate five
models and theories (ISS, TAM, affinity theory, coolness theory, and
posthumanism) and other important constructs (user ethical perceptions
and user ethical beliefs) in a new model to examine the determinants of
users’ satisfaction, loyalty, and ethical usage concerns of the AI chatbot.
Second, we have demonstrated that the multidimensional construct
of IQ positively impacts perceived usefulness and satisfaction. While the
link between IQ, perceived usefulness, and satisfaction has been widely
discussed in the literature on chatbots, most research has adopted a
holistic approach to IQ without explaining how the underlying mecha­
nisms of IQ impact the outcome variables (e.g., Ashfaq et al., 2020; Ruan
and Mezei, 2022; Tisland et al., 2022). Ge et al. (2011) stressed that IQ is
a complex phenomenon that encompasses important relationships
among multiple factors. By demonstrating that relevance, reliability,
conciseness, and quickness impact perceived usefulness and satisfaction,
this research sheds light on the key aspects of IQ that are important to
customers, providing a framework for companies to develop effective Fig. 2. Interaction effect SATxUEP→UL.

Table 3
Discriminant validity analysis.
Constructs 1 2 3 4 5 6 7 8 9 10 11

Relevance of Information 0.843 0.555 0.557 0.479 0.596 0.593 0.541 0.534 0.632 0.096 0.104
Reliability of Information 0.556 0.858 0.52 0.415 0.587 0.578 0.517 0.51 0.553 0.155 0.104
Conciseness of Information 0.562 0.518 0.865 0.39 0.54 0.508 0.501 0.501 0.577 0.163 0.083
Quickness of information 0.473 0.412 0.39 0.891 0.454 0.462 0.467 0.423 0.452 0.181 0.178
Perceived usefulness 0.598 0.587 0.54 0.456 0.854 0.59 0.553 0.558 0.628 0.164 0.11
perceived coolness 0.593 0.578 0.509 0.465 0.591 0.852 0.534 0.544 0.616 0.135 0.158
Satisfaction 0.539 0.517 0.5 0.463 0.554 0.537 0.869 0.534 0.541 0.136 0.05
Technology affinity 0.534 0.51 0.501 0.423 0.56 0.543 0.534 0.844 0.557 0.09 0.141
User loyalty 0.631 0.553 0.578 0.453 0.627 0.618 0.54 0.558 0.844 0.138 0.167
User ethical perceptions 0.094 0.153 0.162 − 0.182 0.165 0.135 − 0.137 0.093 0.139 0.786 0.117
User ethical beliefs 0.101 0.098 0.082 0.182 0.107 0.156 − 0.05 0.14 0.168 − 0.116 0.779

Note: SD = standard deviation; bold values = square root of the average variance extracted (AVE); off-diagonals = correlations.

8
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

relevant information. Reliability—the AI chatbot should be accurate and

consistent in its responses. To ensure reliability, the chatbot should be
trained on high-quality data sources and regularly tested for accuracy
and consistency. Conciseness—the virtual agent should provide the
necessary information in a clear and concise manner, without unnec­
essary detail. To ensure conciseness, programmers should conduct reg­
ular assessments of the chatbot’s language and tone to ensure it is clear
and concise. Quickness—the chatbot should respond quickly to user
queries without unnecessary delay. To ensure quickness, programmers
should monitor chatbot response times to identify areas where the
chatbot is slow and take steps to address these issues. Additionally, they
should optimize the chatbot’s infrastructure and backend processes to
ensure fast and efficient responses. In sum, brand managers should
monitor chatbot performance metrics to identify areas where the chat­
bot is not providing relevant, reliable, and concise information and take
steps to address these issues.
Second, although we found that IQ is one of the most important
Fig. 3. Interaction effect SATxUEB→UL. determinants of users’ satisfaction and loyalty toward ChatGPT, the
strong association between perceived coolness and users’ satisfaction
also provides a strong impetus for ChatGPT’s parent company, OpenAI,
strategies for improving the quality of information.
and other companies building chatbots based on transformer technology
Third, despite the extensive research conducted on chatbots in recent
to design cool chatbots and develop a cool image around them. As the
years, the predictors of user satisfaction and loyalty are still limited. This
use of these types of chatbots increases, brands may find it challenging
study brings coolness theory to the forefront of research on chatbots by
to retain customer loyalty. Therefore, they could successfully design
demonstrating that the quality of the information delivered by chatbots
unique chatbots equipped with cool features and functions using the
can be perceived as cool by users when the information is quick, rele­
coolness framework, enhancing users’ satisfaction and loyalty. As noted
vant, reliable, and concise. Additionally, we have confirmed that
by Sundar et al. (2014), what is cool today may not be cool tomorrow.
perceived coolness positively impacts user satisfaction. This finding
This implies that these types of chatbots may not be considered cool
aligns with earlier research suggesting that perceived coolness is a
when they are fully adopted. Thus, brands should strive to develop new
critical predictor of user satisfaction with a technological device (Liu
features regularly to make their virtual agents stand apart from other
and Mattila, 2019; Nan et al., 2022b).
chatbots, create new stimuli, or leverage existing ones while building
Fourth, we have found that technology affinity is a significant factor
communication strategies.
in determining user satisfaction. Previous studies have already investi­
Third, as evidenced by the findings, ChatGPT’s posthuman ability
gated the association between technology affinity and satisfaction. For
also influences user satisfaction and loyalty. This suggests an important
example, Xu and Du (2018) examined the impact of digital libraries’
consideration for designers of these technologies: they should not use
affinity on enhancing user satisfaction. What sets our research apart is
human abilities as the benchmark for their features and functions.
that we demonstrate that users can develop an affinity for chatbots,
Instead, they should strive to explore ways in which technology can
which ultimately influences their satisfaction.
exceed human capabilities. Describing these technologies solely in terms
Fifth, we have shown that ChatGPT’s posthuman ability positively
of human roles can be limiting for users. For instance, just as calculators
impacts users’ satisfaction, suggesting that digital devices that signifi­
and computers are not marketed as approximating the function of a
cantly outperform humans or assist them in completing tasks efficiently
human accountant, the capabilities of ChatGPT, such as “proofreading
can be perceived as posthuman and satisfy users’ needs. Previous
and editing,” should not be overemphasized. Rather, the focus should be
research on posthuman ability has primarily focused on investigating its
on the overall usefulness of the technology in meeting users’ needs.
association with technology acceptance (e.g., Gambino and Sundar,
One of the key findings of this research is that user ethical percep­
2019). Thus, our study is unique in establishing a relationship between
tions of ChatGPT and user ethical beliefs have a negative moderating
posthuman ability and satisfaction, adding to the existing literature on
effect on the association between satisfaction and loyalty. Therefore,
this topic.
chatbot brands need to recognize the crucial role of users’ ethical per­
Finally, we have confirmed that user ethical perceptions and beliefs
ceptions and beliefs while interacting with chatbots. They should assure
moderate the association between user satisfaction and loyalty. To the
users of their privacy and implement robust data protection measures to
best of our knowledge, no previous research has explored the logical
help build trust and alleviate concerns about the ethical implications of
possibility of user ethical perceptions and user ethical beliefs moder­
using AI chatbots. Demonstrating a commitment to user data privacy
ating the satisfaction-loyalty relationship. Therefore, our study’s find­
and security can encourage users to continue using the system and in­
ings are noteworthy as they provide evidence for this possibility.
crease their loyalty. Additionally, they should establish and publicize
ethical guidelines and policies to demonstrate the organization’s
8.2. Managerial implications commitment to responsible AI use. These guidelines should address
concerns such as bias, stereotypes, discrimination, and the potential
The present study highlights several key implications for pro­ impact of AI on society. By proactively addressing ethical concerns,
grammers and brand managers seeking to promote the healthy devel­ managers can mitigate negative perceptions and enhance user loyalty.
opment of chatbots, including ChatGPT specifically.
First, based on our findings, we recommend that brand managers 8.3. Limitations and future research lines
focus on the four factors of IQ (i.e., relevance, reliability, conciseness,
and quickness) to enhance users’ perceptions of usefulness, coolness, Like any other study, this research has several limitations that war­
and satisfaction with ChatGPT. Relevance—the chatbot must be able to rant scholarly attention. First, we employed a non-probability purposive
understand the user’s needs and provide accurate and useful responses. sampling method, which raises concerns regarding the sample’s repre­
To ensure relevance, brand managers should conduct regular assess­ sentativeness. Furthermore, the utilization of self-administered online
ments of user needs and feedback to ensure the chatbot is providing questionnaires excludes certain demographic groups, particularly older

9
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

individuals and those with lower academic qualifications. This exclusion Data availability
may affect the overall representativeness of the sample. Future studies
could employ a random sampling technique to investigate the theoret­ Data will be made available on request.
ical framework of this study on a more diverse and inclusive sample of
users. References
Secondly, it is important to acknowledge that our data was exclu­
sively collected from participants in the US. As a result, the study pro­ Ahn, H., Park, E., 2022. Determinants of consumer acceptance of mobile healthcare
devices: an application of the concepts of technology acceptance and coolness.
vides a relatively narrow perspective on the global phenomenon of Telematics Inf. 70, 101810 https://doi.org/10.1016/j.tele.2022.101810.
ChatGPT, lacking international, cross-cultural, and global viewpoints Ajzen, I., Fishbein, M., 2005. The infuence of attitudes on behavior. In: Albarracín, D.,
that could enhance our understanding in a more comprehensive manner. Johnson, B.T., Zanna, M.P. (Eds.), The Handbook of Attitudes. Lawrence Erlbaum
Associates Publishers, Mahwah, NJ, pp. 173–221.
Future research could conduct cross-country studies or validate the Al-Fraihat, D., Joy, M., Sinclair, J., 2020. Evaluating E-learning systems success: an
proposed model in other countries, such as China, to further broaden the empirical study. Comput. Hum. Behav. 102, 67–86. https://doi.org/10.1016/j.
scope of our investigation. chb.2019.08.004.
Aldás-Manzano, J., Ruiz-Mafé, C., Sanz-Blas, S., 2009. Exploring individual personality
Third, we solely focused on satisfaction as a mediator in our model. factors as drivers of M-shopping acceptance. Ind. Manag. Data Syst. 109 (6),
Future studies could enhance the model by incorporating an emotional 739–757. https://doi.org/10.1108/02635570910968018.
attitude, such as affective commitment, to ascertain the degree to which Araujo, T., 2018. Living up to the chatbot hype: the influence of anthropomorphic design
cues and communicative agency framing on conversational agent and company
both emotional and cognitive attitudes amplify the influence of the
perceptions. Comput. Hum. Behav. 85, 183–189. https://doi.org/10.1016/j.
predictor variables on the outcome variable. Fourth, this research did chb.2018.03.051.
not consider the age categories and genders of the respondents. Future Arghashi, V., Yuksel, C.A., 2022. Interactivity, Inspiration, and Perceived Usefulness!
research could investigate whether different age groups and genders How retailers’ AR-apps improve consumer engagement through flow. J. Retailing
Consum. Serv. 64, 102756 https://doi.org/10.1016/j.jretconser.2021.102756.
hold different attitudes toward ChatGPT, which may potentially influ­ Arif, T. Bin, Munaf, U., Ul-Haque, I., 2023. The future of medical education and research:
ence their behaviors. is ChatGPT a blessing or blight in disguise? Med. Educ. Online 28 (Issue 1), 2181052.
Finally, a recent study on 1000 US students found that one-third had https://doi.org/10.1080/10872981.2023.2181052. Taylor & Francis.
Ashfaq, M., Yun, J., Yu, S., Loureiro, S.M.C., 2020. I, Chatbot: modeling the determinants
used ChatGPT to complete their academic assignments (Intelligent, of users’ satisfaction and continuance intention of AI-powered service agents.
2023), raising concerns about academic dishonesty in the use of AI Telematics Inf. 54, 101473 https://doi.org/10.1016/j.tele.2020.101473.
chatbots. Future studies could explore the factors that could assist aca­ Ben-Shabat, N., Sharvit, G., Meimis, B., Joya, D. Ben, Sloma, A., Kiderman, D., Shabat, A.,
Tsur, A.M., Watad, A., Amital, H., 2022. Assessing data gathering of chatbot based
demic institutions in preventing students’ unethical use of AI chatbots. symptom checkers-a clinical vignettes study. Int. J. Med. Inf. 168, 104897 https://
doi.org/10.1016/j.ijmedinf.2022.104897.
9. Conclusion Bloemer, J., De Ruyter, K.O., Wetzels, M., 1999. Linking perceived service quality and
service loyalty: a multidimensional perspective. Eur. J. Market. 33 (11/12),
1082–1106. https://doi.org/10.1108/03090569910292285.
In conclusion, this research provides valuable insights into the fac­ Bostrom, N., 2013. Why I Want to Be a Posthuman when I Grow up. The Transhumanist
tors influencing users’ satisfaction and loyalty towards ChatGPT. The Reader: Classical and Contemporary Essays on the Science, Technology, and
Philosophy of the Human Future, pp. 28–53. https://doi.org/10.1002/
findings highlight the significance of several factors, including IQ,
9781118555927.ch3.
perceived usefulness, technology affinity, perceived coolness, and Celik, I., 2023. Towards Intelligent-TPACK: an empirical study on teachers’ professional
posthuman ability, in shaping users’ attitudes and behaviors. The study knowledge to ethically integrate artificial intelligence (AI)-based tools into
emphasizes the importance of delivering relevant, reliable, concise, and education. Comput. Hum. Behav. 138, 107468 https://doi.org/10.1016/j.
chb.2022.107468.
prompt information to enhance users’ perception of usefulness, cool­ Chaudhry, P.E., Stumpf, S.A., 2011. Consumer complicity with counterfeit products.
ness, and satisfaction. Furthermore, factors such as perceived usefulness, J. Consum. Market. 28 (2), 139–151. https://doi.org/10.1108/
technology affinity, perceived coolness, and the perception of advanced 07363761111115980.
Chen, C.-F., Chou, S.-H., 2019. Antecedents and consequences of perceived coolness for
capabilities (posthuman ability) all contribute to users’ satisfaction and Generation Y in the context of creative tourism-A case study of the Pier 2 Art Center
loyalty. However, it is crucial to address users’ ethical concerns as they in Taiwan. Tourism Manag. 72, 121–129. https://doi.org/10.1016/j.
can have a negative impact on their loyalty toward AI chatbots. Overall, tourman.2018.11.016.
Chen, C.-C., Tsai, J.-L., 2019. Determinants of behavioral intention to use the
these findings provide valuable guidance for programmers and AI firms Personalized Location-based Mobile Tourism Application: an empirical study by
aiming to enhance user satisfaction, foster loyalty, and address ethical integrating TAM with ISSM. Future Generat. Comput. Syst. 96, 628–638. https://doi.
considerations in the development and deployment of generative org/10.1016/j.future.2017.02.028.
Cheng, X., Gu, Y., Hua, Y., Luo, X.R., 2021. The paradox of word-of-mouth in social
conversational AI agents like ChatGPT. commerce: exploring the juxtaposed impacts of source credibility and information
quality on SWOM spreading. Inf. Manag. 58 (7), 103505 https://doi.org/10.1016/j.
Funding im.2021.103505.
Cheung, M.F.Y., To, W.M., 2021. The effect of consumer perceptions of the ethics of
retailers on purchase behavior and word-of-mouth: the moderating role of ethical
The study is supported by Key Program of National Natural Science beliefs. J. Bus. Ethics 171 (4), 771–788. https://doi.org/10.1007/s10551-020-
Foundation of China (Nos.72334004), General Program of National 04431-6.
Natural Science Foundation of China (Nos.71971143, 72374143), Chiu, C.-M., Cheng, H.-L., Huang, H.-Y., Chen, C.-F., 2013. Exploring individuals’
subjective well-being and loyalty towards social network sites from the perspective
Guangdong Provincial Philosophy and Social Sciences Planning Project of network externalities: the Facebook case. Int. J. Inf. Manag. 33 (3), 539–552.
(No. GD22CGL35), Special Projects in Key Fields of Ordinary Colleges https://doi.org/10.1016/j.ijinfomgt.2013.01.007.
and Universities in Guangdong Province (No. 2022ZDZX2054), and Chung, M., Ko, E., Joung, H., Kim, S.J., 2020. Chatbot e-service and customer satisfaction
regarding luxury brands. J. Bus. Res. 117, 587–595. https://doi.org/10.1016/j.
University Innovation Team Project of Guangdong Province (No. jbusres.2018.10.004.
2021WCXTD002). Dai, H., Liu, Z., Liao, W., Huang, X., Wu, Z., Zhao, L., Liu, W., Liu, N., Li, S., Zhu, D.,
2023. Chataug: Leveraging Chatgpt for Text Data Augmentation. ArXiv Preprint
ArXiv:2302.13007.
Declaration of competing interest Davis, F.D., 1985. A Technology Acceptance Model for Empirically Testing New End-
User Information Systems: Theory and Results. Massachusetts Institute of
The authors declare that they have no known competing financial Technology.
Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of
interests or personal relationships that could have appeared to influence information technology. MIS Q. 13 (3), 319–340.
the work reported in this paper. DeLone, W.H., McLean, E.R., 1992. Information systems success: the quest for the
dependent variable. Inf. Syst. Res. 3 (1), 60–95. https://doi.org/10.1287/isre.3.1.60.

10
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

Demoulin, N.T.M., Coussement, K., 2020. Acceptance of text-mining systems: the content management system users. Eur. J. Inf. Syst. 26 (4), 333–360. https://doi.
signaling role of information quality. Inf. Manag. 57 (1), 103120 https://doi.org/ org/10.1057/s41303-016-0029-7.
10.1016/j.im.2018.10.006. Li, C.-Y., Fang, Y.-H., 2019. Predicting continuance intention toward mobile branded
Dwivedi, Y.K., Kshetri, N., Hughes, L., Slade, E.L., Jeyaraj, A., Kar, A.K., Baabdullah, A. apps through satisfaction and attachment. Telematics Inf. 43, 101248 https://doi.
M., Koohang, A., Raghavan, V., Ahuja, M., 2023. “So what if ChatGPT wrote it?” org/10.1016/j.tele.2019.101248.
Multidisciplinary perspectives on opportunities, challenges and implications of Liu, S.Q., Mattila, A.S., 2019. Apple Pay: coolness and embarrassment in the service
generative conversational AI for research, practice and policy. Int. J. Inf. Manag. 71, encounter. Int. J. Hospit. Manag. 78, 268–275. https://doi.org/10.1016/j.
102642 https://doi.org/10.1016/j.ijinfomgt.2023.102642. ijhm.2018.09.009.
Eysenbach, G., 2023. The role of chatgpt, generative language models, and artificial Liu, C.-T., Guo, Y.M., Lee, C.-H., 2011. The effects of relationship quality and switching
intelligence in medical education: a conversation with chatgpt and a call for papers. barriers on customer loyalty. Int. J. Inf. Manag. 31 (1), 71–79. https://doi.org/
JMIR Med. Educat. 9 (1), e46885 https://doi.org/10.2196/46885. 10.1016/j.ijinfomgt.2010.05.008.
Focus, Science, 2023. ChatGPT: Everything You Need to Know about OpenAI’s GPT-3 Loeppky, J., 2023. ChatGPT vs Chatbots: What’s the Difference? https://www.itpro.co
Tool. https://www.sciencefocus.com/future-technology/gpt-3/. (Accessed 19 May m/technology/artificial-intelligence-ai/369979/chatgpt-vs-chatbots-whats-the-diffe
2023). rence.
Gambino, A., Sundar, S.S., 2019. Acceptance of self-driving cars: does their posthuman Ma, Y., 2021. Elucidating determinants of customer satisfaction with live-stream
ability make them more eerie or more desirable?. In: Extended Abstracts of the 2019 shopping: an extension of the information systems success model. Telematics Inf. 65,
CHI Conference on Human Factors in Computing Systems, pp. 1–6. https://doi.org/ 101707 https://doi.org/10.1016/j.tele.2021.101707.
10.1145/3290607.3312870. Mamonov, S., Koufaris, M., 2020. Fulfillment of higher-order psychological needs
Ge, Mouzhi, Helfert, Markus, Jannach, Dietmar, 2011. Information Quality Assessment: through technology: the case of smart thermostats. Int. J. Inf. Manag. 52, 102091
Validating Measurement Dimensions and Processes. ECIS 2011 Proceedings, 75. https://doi.org/10.1016/j.ijinfomgt.2020.102091.
https://aisel.aisnet.org/ecis2011/75. McGowan, L., 2023. OpenAI Criticised for Lack of Transparency Around GPT-4. https
Germain, T., 2023. They’re All So Dirty and Smelly:’ Study Unlocks ChatGPT’s Inner ://www.siliconrepublic.com/machines/openai-gpt4-transparency-ai-concerns-strip
Racist. Available at: (https://gizmodo.com/chatgpt-ai-openai-study-frees-chat-gpt e-chatgpt.
-inner-racist-1850333646. McKinsey, 2022. Generative AI Is Here: How Tools like ChatGPT Could Change Your
Ghasemaghaei, M., Hassanein, K., 2015. Online information quality and consumer Business. https://www.mckinsey.com/capabilities/quantumblack/our-insights/ge
satisfaction: the moderating roles of contextual factors–A meta-analysis. Inf. Manag. nerative-ai-is-here-how-tools-like-chatgpt-could-change-your-business (Accessed 15
52 (8), 965–981. https://doi.org/10.1016/j.im.2015.07.001. July 2023).
Gkinko, L., Elbanna, A., 2022. The appropriation of conversational AI in the workplace: a Milmo, D., 2023. ChatGPT Reaches 100 Million Users Two Months after Launch. https
taxonomy of AI chatbot users. Int. J. Inf. Manag. 102568 https://doi.org/10.1016/j. ://www.theguardian.com/technology/2023/feb/02/chatgpt-100-million-users-ope
ijinfomgt.2022.102568. n-ai-fastest-growing-app.
Hausvik, G.I., Thapa, D., Munkvold, B.E., 2021. Information quality life cycle in Mou, Y., Xu, K., 2017. The media inequality: comparing the initial human-human and
secondary use of EHR data. Int. J. Inf. Manag. 56, 102227 https://doi.org/10.1016/ human-AI social interactions. Comput. Hum. Behav. 72, 432–440. https://doi.org/
j.ijinfomgt.2020.102227. 10.1016/j.chb.2017.02.067.
Hendriana, E., Mayasari, A.P., Gunadi, W., 2013. Why do college students buy Mvondo, G.F.N., Jing, F., Hussain, K., Jin, S., Raza, M.A., 2022a. Impact of international
counterfeit movies? Int. J. Educ. Bus. Manag. Learn. 3 (1), 62. tourists’ co-creation experience on brand trust, brand passion, and brand
Hill, J., Ford, W.R., Farreras, I.G., 2015. Real conversations with artificial intelligence: a evangelism. Front. Psychol. 13 (13), 866362 https://doi.org/10.3389/
comparison between human–human online conversations and human–chatbot fpsyg.2022.866362.
conversations. Comput. Hum. Behav. 49, 245–250. https://doi.org/10.1016/j. Mvondo, G.F.N., Jing, F., Hussain, K., Raza, M.A., 2022b. Converting tourists into
chb.2015.02.026. evangelists: exploring the role of tourists’ participation in value co-creation in
Hsu, C.-L., Lin, J.C.-C., 2023. Understanding the user satisfaction and loyalty of customer enhancing brand evangelism, empowerment, and commitment. J. Hospit. Tourism
service chatbots. J. Retailing Consum. Serv. 71, 103211 https://doi.org/10.1016/j. Manag. 52, 1–12. https://doi.org/10.1016/j.jhtm.2022.05.015.
jretconser.2022.103211. Mvondo, G.F.N., Niu, B., Eivazinezhad, S., 2023a. Generative Conversational AI and
Huang, S.Y.B., Lee, C.-J., 2022. Predicting continuance intention to fintech chatbot. Academic Integrity: A Mixed Method Investigation to Understand the Ethical Use of
Comput. Hum. Behav. 129, 107027 https://doi.org/10.1016/j.chb.2021.107027. LLM Chatbots in Higher Education (August 22, 2023). https://doi.org/10.2139/
Indeed, 2022. https://uk.indeed.com/career-advice/career-development/ethical-issue ssrn.4548263. https://ssrn.com/abstract=4548263.
s-in-business. (Accessed 15 July 2023). Mvondo, G.F.N., Jing, F., Hussain, K., 2023b. What’s in the box? Investigating the
Intelligent, 2023. Nearly 1 in 3 College Students Have Used ChatGPT on Written benefits and risks of the blind box selling strategy. J. Retailing Consum. Serv. 71,
Assigments. https://www.intelligent.com/nearly-1-in-3-college-students-have-use 103189 https://doi.org/10.1016/j.jretconser.2022.103189.
d-chatgpt-on-written-assignments/. (Accessed 15 July 2023). Nadeem, W., Juntunen, M., Shirazi, F., Hajli, N., 2020. Consumers’ value co-creation in
Jin, S.V., Youn, S., 2021. Why do consumers with social phobia prefer anthropomorphic sharing economy: the role of social support, consumers’ ethical perceptions and
customer service chatbots? Evolutionary explanations of the moderating roles of relationship quality. Technol. Forecast. Soc. Change 151, 119786. https://doi.org/
social phobia. Telematics Inf. 62, 101644 https://doi.org/10.1016/j. 10.1016/j.techfore.2019.119786.
tele.2021.101644. Nan, D., Lee, H., Kim, Y., Kim, J.H., 2022a. My video game console is so cool! A coolness
Jones, T.O., Sasser Jr., W.E., 1995. Why satisfied customers defect. Harv. Bus. Rev. 73 theory-based model for intention to use video game consoles. Technol. Forecast.
(6), 88e99. Soc. Change 176, 121451. https://doi.org/10.1016/j.techfore.2021.121451.
Joo, Y.J., Park, S., Shin, E.K., 2017. Students’ expectation, satisfaction, and continuance Nan, D., Shin, E., Barnett, G.A., Cheah, S., Kim, J.H., 2022b. Will coolness factors predict
intention to use digital textbooks. Comput. Hum. Behav. 69, 83–90. https://doi.org/ user satisfaction and loyalty? Evidence from an artificial neural network–structural
10.1016/j.chb.2016.12.025. equation model approach. Inf. Process. Manag. 59 (6), 103108 https://doi.org/
Kasilingam, D.L., 2020. Understanding the attitude and intention to use smartphone 10.1016/j.ipm.2022.103108.
chatbots for shopping. Technol. Soc. 62, 101280 https://doi.org/10.1016/j. Nass, C., Steuer, J., Tauber, E.R., 1994. Computers are social actors. Proc. SIGCHI Conf.
techsoc.2020.101280. Human Factors Comput. Syst. 72–78.
Kim, T.B., Ho, C.-T.B., 2021. Validating the moderating role of age in multi-perspective Nath, R., Manna, R., 2023. From posthumanism to ethics of artificial intelligence. AI Soc.
acceptance model of wearable healthcare technology. Telematics Inf. 61, 101603 38 (1), 185–196. https://doi.org/10.1007/s00146-021-01274-1.
https://doi.org/10.1016/j.tele.2021.101603. Nguyen, Q.N., Sidorova, A., Torres, R., 2022. User interactions with chatbot interfaces vs.
Kim, C., Oh, E., Shin, N., Chae, M., 2009. An empirical investigation of factors affecting Menu-based interfaces: an empirical study. Comput. Hum. Behav. 128, 107093
ubiquitous computing use and U-business value. Int. J. Inf. Manag. 29 (6), 436–448. https://doi.org/10.1016/j.chb.2021.107093.
https://doi.org/10.1016/j.ijinfomgt.2009.06.003. Oliver, R.L., 1980. A cognitive model of the antecedents and consequences of satisfaction
Kock, N., 2015. Common method bias in PLS-SEM: a full collinearity assessment decisions. J. Market. Res. 17 (4), 460–469. https://doi.org/10.1177/
approach. Int. J. e-Collaboration 11 (4), 1–10. https://doi.org/10.4018/ 002224378001700405.
ijec.2015100101. Oliver, R.L., 1999. Whence consumer loyalty? J. Market. 63 (4_Suppl. 1), 33–44. https://
Konya-Baumbach, E., Biller, M., von Janda, S., 2023. Someone out there? A study on the doi.org/10.1177/00222429990634s105.
social presence of anthropomorphized chatbots. Comput. Hum. Behav. 139, 107513 OpenAI, 2022. ChatGPT: Optimizing Language Models for Dialogue. https://openai.
https://doi.org/10.1016/j.chb.2022.107513. com/blog/chatgpt/. (Accessed 15 July 2023).
Kooli, C., 2023. Chatbots in education and research: a critical examination of ethical OpenAI, 2023. What Is ChatGPT. https://help.openai.com/en/articles/6783457-what
implications and solutions. Sustainability 15 (7), 5614. https://doi.org/10.3390/ -is-chatgpt. (Accessed 15 July 2023).
su15075614. Peng, X., Zhao, Y.C., Teo, H.H., 2016. Understanding young people’s use of Danmaku
Kumar, P., 2023. Effects of in-store information quality and store credibility on consumer websites: the effect of perceived coolness and subcultural identi. PACIS 2016
engagement in green retailing. J. Retailing Consum. Serv. 71, 103195 https://doi. Proceedings 252. http://aisel.aisnet.org/pacis2016/252.
org/10.1016/j.jretconser.2022.103195. Perse, E.M., 1986. Soap opera viewing patterns of college students and cultivation.
Kwangsawad, A., Jattamart, A., 2022. Overcoming customer innovation resistance to the J. Broadcast. Electron. Media 30 (2), 175–193.
sustainable adoption of chatbot services: a community-enterprise perspective in Pop, R.-A., Hlédik, E., Dabija, D.-C., 2023. Predicting consumers’ purchase intention
Thailand. J. Innovat. Knowl. 7 (3), 100211 https://doi.org/10.1016/j. through fast fashion mobile apps: the mediating role of attitude and the moderating
jik.2022.100211. role of COVID-19. Technol. Forecast. Soc. Change 186, 122111. https://doi.org/
Laumer, S., Maier, C., Weitzel, T., 2017. Information quality, user satisfaction, and the 10.1016/j.techfore.2022.122111.
manifestation of workarounds: a qualitative and quantitative study of enterprise

11
B. Niu and G.F.N. Mvondo Journal of Retailing and Consumer Services 76 (2024) 103562

Rajaobelina, L., Prom Tep, S., Arcand, M., Ricard, L., 2021. Creepiness: its antecedents of using chatbots in education. Smart Learn. Environ. 10 (1), 15. https://doi.org/
and impact on loyalty when interacting with a chatbot. Psychol. Market. 38 (12), 10.1186/s40561-023-00237-x.
2339–2356. https://doi.org/10.1002/mar.21548. Torres, R., Sidorova, A., 2019. Reconceptualizing information quality as effective use in
Reinikainen, H., Tan, T.M., Luoma-aho, V., Salo, J., 2021. Making and breaking the context of business intelligence and analytics. Int. J. Inf. Manag. 49, 316–329.
relationships on social media: the impacts of brand and influencer betrayals. https://doi.org/10.1016/j.ijinfomgt.2019.05.028.
Technol. Forecast. Soc. Change 171, 120990. https://doi.org/10.1016/j. Tran, A.D., Pallant, J.I., Johnson, L.W., 2021. Exploring the impact of chatbots on
techfore.2021.120990. consumer sentiment and expectations in retail. J. Retailing Consum. Serv. 63,
Román, S., 2007. The ethics of online retailing: a scale development and validation from 102718 https://doi.org/10.1016/j.jretconser.2021.102718.
the consumers’ perspective. J. Bus. Ethics 72, 131–148. https://doi.org/10.1007/ Van den Broeck, E., Zarouali, B., Poels, K., 2019. Chatbot advertising effectiveness: when
s10551-006-9161-y. does the message get through? Comput. Hum. Behav. 98, 150–157. https://doi.org/
Ruan, Y., Mezei, J., 2022. When do AI chatbots lead to higher customer satisfaction than 10.1016/j.chb.2019.04.009.
human frontline employees in online shopping assistance? Considering product Veale, K., 2023. Does ChatGPT Have Privacy Issues? https://www.makeuseof.com/chat
attribute type. J. Retailing Consum. Serv. 68, 103059 https://doi.org/10.1016/j. gpt-privacy-issues/.
jretconser.2022.103059. Verma, S., Bhattacharyya, S.S., Kumar, S., 2018. An extension of the technology
Rubin, A.M., 1981. An examination of television viewing motivations. Commun. Res. 8 acceptance model in the big data analytics system implementation environment. Inf.
(2), 141–165. https://doi.org/10.1177/009365028100800201. Process. Manag. 54 (5), 791–806. https://doi.org/10.1016/j.ipm.2018.01.004.
Sarkar, S., Chauhan, S., Khare, A., 2020. A meta-analysis of antecedents and Vitell, S.J., Paolillo, J.G., 2004. A cross-cultural study of the antecedents of the perceived
consequences of trust in mobile commerce. Int. J. Inf. Manag. 50, 286–301. https:// role of ethics and social responsibility. Bus. Ethics Eur. Rev. 13 (2–3), 185–199.
doi.org/10.1016/j.ijinfomgt.2019.08.008. Wang, E.S.-T., Lin, R.-L., 2017. Perceived quality factors of location-based apps on trust,
Selamat, M.A., Windasari, N.A., 2021. Chatbot for SMEs: integrating customer and perceived privacy risk, and continuous usage intention. Behavior Inform. Technol.
business owner perspectives. Technol. Soc. 66, 101685 https://doi.org/10.1016/j. 36 (1), 2–10. https://doi.org/10.1080/0144929X.2016.1143033.
techsoc.2021.101685. Wang, R.Y., Strong, D.M., 1996. Beyond accuracy: what data quality means to data
Seong, B.-H., Hong, C.-Y., 2022. Corroborating the effect of positive technology readiness consumers. J. Manag. Inf. Syst. 12 (4), 5–33. https://doi.org/10.1080/
on the intention to use the virtual reality sports game "Screen Golf": focusing on the 07421222.1996.11518099.
technology readiness and acceptance model. Inf. Process. Manag. 59 (4), 102994 Wang, C., Li, Y., Fu, W., Jin, J., 2023. Whether to trust chatbots: applying the event-
https://doi.org/10.1016/j.ipm.2022.102994. related approach to understand consumers’ emotional experiences in interactions
Setia, P., Setia, P., Venkatesh, V., Joglekar, S., 2013. Leveraging digital technologies: with chatbots in e-commerce. J. Retailing Consum. Serv. 73, 103325 https://doi.
how information quality leads to localized capabilities and customer service org/10.1016/j.jretconser.2023.103325.
performance. MIS Q. 565–590. Warren, C., Pezzuti, T., Koley, S., 2018. Is being emotionally inexpressive cool?
Shawar, B.A., Atwell, E.S., 2005. Using corpora in machine-learning chatbot systems. Int. J. Consum. Psychol. 28 (4), 560–577. https://doi.org/10.1002/jcpy.1039.
J. Corpus Linguist. 10 (4), 489–516. https://doi.org/10.1075/ijcl.10.4.06sha. Xu, F., Du, J.T., 2018. Factors influencing users’ satisfaction and loyalty to digital
Similarweb, 2023. ChatGPT Tops 25 Million Daily Visits. https://www.similarweb.com/ libraries in Chinese universities. Comput. Hum. Behav. 83, 64–72. https://doi.org/
blog/insights/chatgpt-25-million/. (Accessed 8 May 2023). 10.1016/j.chb.2018.01.029.
Smith, C., 2023. Bing Chat Ads Will Start the ChatGPT Privacy Discussion We Xu, C., Peak, D., Prybutok, V., 2015. A customer value, satisfaction, and loyalty
Desperately Need. https://bgr.com/tech/ads-in-bing-chat-will-open-the-chatgpt-pri perspective of mobile application recommendations. Decis. Support Syst. 79,
vacy-talk-we-desperately-need-and-i-cant-wait/. 171–183. https://doi.org/10.1016/j.dss.2015.08.008.
Soltani-Nejad, N., Taheri-Azad, F., Zarei-Maram, N., Saberi, M.K., 2020. Developing a Zhao, L., Lu, Y., Zhang, L., Chau, P.Y.K., 2012. Assessing the effects of service quality and
model to identify the antecedents and consequences of user satisfaction with digital justice on customer satisfaction and the continuance intention of mobile value-added
libraries. Aslib J. Inf. Manag. 72 (6), 979–997. https://doi.org/10.1108/AJIM-04- services: an empirical test of a multidimensional model. Decis. Support Syst. 52 (3),
2020-0099. 645–656. https://doi.org/10.1016/j.dss.2011.10.022.
Song, M., Xing, X., Duan, Y., Cohen, J., Mou, J., 2022. Will artificial intelligence replace Zhong, J., Chen, T., 2023. Antecedents of mobile payment loyalty: an extended
human customer service? The impact of communication quality and privacy risks on perspective of perceived value and information system success model. J. Retailing
adoption intention. J. Retailing Consum. Serv. 66, 102900 https://doi.org/10.1016/ Consum. Serv. 72, 103267 https://doi.org/10.1016/j.jretconser.2023.103267.
j.jretconser.2021.102900. Zhou, Q., Li, B., Han, L., Jou, M., 2023. Talking to a bot or a wall? How chatbots vs.
Sundar, S.S., Tamul, D.J., Wu, M., 2014. Capturing "cool": measures for assessing human agents affect anticipated communication quality. Comput. Hum. Behav.,
coolness of technological products. Int. J. Hum. Comput. Stud. 72 (2), 169–180. 107674 https://doi.org/10.1016/j.chb.2023.107674.
https://doi.org/10.1016/j.ijhcs.2013.09.008. Zrnec, A., Poženel, M., Lavbič, D., 2022. Users’ ability to perceive misinformation: an
Teo, T.S.H., Srivastava, S.C., Jiang, L.I., 2008. Trust and electronic government success: information quality assessment approach. Inf. Process. Manag. 59 (1), 102739
an empirical study. J. Manag. Inf. Syst. 25 (3), 99–132. https://doi.org/10.2753/ https://doi.org/10.1016/j.ipm.2021.102739.
MIS0742-1222250303.
Thakur, R., 2018. The role of self-efficacy and customer satisfaction in driving loyalty to
Prof. Ben Niu: He is the Professor and Head of Management Department at Shenzhen
the mobile shopping application. Int. J. Retail Distrib. Manag. 46 (3), 283–303.
University, China. He is a Seasoned Researcher with a good number of research publica­
https://doi.org/10.1108/IJRDM-11-2016-0214.
tions. His research appeared in well reputed journals including IEEE Transactions on Evo­
Tisland, I., Sodefjed, M.L., Vassilakopoulou, P., Pappas, I.O., 2022. The role of quality,
lution Computation, IEEE Transactions on Systems Man and Cybernetic, Journal of Cleaner
trust, and empowerment in explaining satisfaction and use of chatbots in e-
Production, and International Journal of Information Management.
government. In: The Role of Digital Technologies in Shaping the Post-Pandemic
World: 21st IFIP WG 6.11 Conference on E-Business, E-Services and E-Society, I3E
2022, pp. 279–291. https://doi.org/10.1007/978-3-031-15342-6_22. Newcastle Dr. Gustave Florentin Nkoulou Mvondo: He is a Postdoctoral Researcher at School of
upon Tyne, UK, September 13–14, 2022, Proceedings. Management, Shenzhen University, China. He has published research articles in peer
Tlili, A., Shehata, B., Adarkwah, M.A., Bozkurt, A., Hickey, D.T., Huang, R., reviewed international journals including Journal of Retailing and Consumer Services,
Agyemang, B., 2023. What if the devil is my guardian angel: ChatGPT as a case study Journal of Hospitality and Tourism Management, and Frontiers in Psychology.

12

View publication stats