2025 Volume 45 Issue 2 Pages 114-123
In the modern world, people have become accustomed to living with various smart objects, such as smartphones, wearable devices, and Internet of Things home appliances equipped with AI. While living with smart objects has many benefits, will such an everyday life lead to a happy future for us? This study examines how two intelligent attributes of AI-enabled smart speakers affect consumer well-being from the perspective of the means-end chain theory. We conducted an online survey of 234 smart voice assistant (SVA) users and analyzed their responses via structural equation modeling. We found that autonomy mainly encouraged consumers to expand themselves, and their interaction experience directly enhanced their subjective well-being toward living with SVAs and indirectly increased it by mediating anthropomorphism and approachability toward SVAs. However, interactivity only encouraged consumers to extend themselves, and this interaction experience directly enhanced their subjective well-being regarding living with SVAs and indirectly increased it only by mediating approachability toward SVAs.
現代では，スマートフォンやウェアラブル端末，AI搭載のIoT家電など，さまざまなスマートオブジェクトが私たちの生活に浸透している。スマートオブジェクトとの共同生活で多くの恩恵がもたらされる一方で，そのような日常が私たちに幸せな未来をもたらすのだろうか。本研究は，AI搭載のスマートスピーカーの2つの主要なインテリジェンス特性が消費者のウェルビーイングに与えるメカニズムを，手段-目的連鎖モデルの視座から検討する。本研究では，SVAsのユーザー234名に対してオンライン調査を行い，構造方程式モデリングによって分析を行った。その結果，自律性は主に消費者の自己拡大を引き起こし，その相互作用経験は直接的にSVAsとの共同生活に対する主観的ウェルビーイングを高め，間接的にもSVAsに対する擬人化と親近感を媒介して高めることがわかった。一方で，双方向性は消費者の自己拡張のみを引き起こし，その相互作用経験は直接的にSVAsとの共同生活に対する主観的ウェルビーイングを高め，間接的にはSVAsに対して親近感のみを媒介して高めることがわかった。
In recent years, the widespread use of Internet of Things (IoT) appliances and other Internet-connected devices has made such tools indispensable in our lives. AI-enabled devices have the ability to connect objects, people, and the physical environment, and are thus termed smart objects (Verhoef et al., 2017). This study focuses on smart voice assistants (SVAs) and smart speakers equipped with AI (Alexa, Google Assistant, and Siri), which are among the most popular smart objects. The global smart speaker market was estimated to be worth USD 12.52 billion in 2023 and is expected to increase from USD 15 billion in 2024 to USD 61.4 billion by 2032, representing a compound annual growth rate of 19.3% over the forecast period (Fortune Business Insight, 2024).
SVAs can suggest music and news that match user preferences, operate air conditioning to keep room temperature constant, adjust lighting according to time of day, and can even engage in light conversation. Living with SVAs brings about improved lifestyles and offers many other benefits. As SVAs have become increasingly common in our daily lives, they have become more familiar. Will such an everyday life lead to a happy future for us? This study examines how the interaction between consumers and SVAs affects consumer well-being.
As SVAs represent a new interface of hands-free and voice-based communication, considerable prior research has examined the interaction between consumers and SVAs, whereby two main focuses have emerged (Kang, Shao, Du, Chen, & Zhang, 2024). One group of studies focuses on the product attributes of SVAs (e.g., Kang, Shao, & Zhang, 2024; Mclean et al., 2021). The second group focuses on user perceptions and psychological states toward SVAs (Hu et al., 2021; Mishra et al., 2022). The current study belongs to the former category, as it focuses on examining how intelligent attributes affect consumer well-being.
These studies commonly apply perspectives such as technology acceptance models developed through research on conventional IT when trying to understand the effects of intelligent attributes. However, although this research starts from intelligent attributes, the outcome is consumer well-being; thus, a more appropriate theoretical background is needed to understand the mechanism. Therefore, this study investigates the interaction between consumers and SVAs based on the means-end chain (MEC) theory.
The MEC theory provides a theoretical basis for the potential relationship between consumer decision-making and cognitive structure. Specifically, it can deepen our understanding of how consumers perceive objects by combining an object’s attributes with a consumer’s goals (Rokeach, 1973). The MEC theory comprises the cognitive levels of attributes, consequences, and values (Gutman, 1982), with a mutual relationship existing among these concepts.
Attributes are the tangible or intangible characteristics of an object that consumers can directly perceive, corresponding to the intelligent attributes of SVAs. Consequences are the results of purchasing, using, and consuming an object, and comprise both functional and psychological consequences. Functional consequences are the functional benefits that consumers obtain from an object, whereas psychological consequences are emotional or social benefits. As discussed below, this study assumes that approachability has an emotional benefit and that anthropomorphism has a social benefit. Values, such as well-being and other perceived ideal states, are abstract and essential outcomes that consumers seek to acquire (Lin et al., 2018).
To understand how intelligent attributes affect consumer well-being, this study hypothesizes two enabling experiences between attributes and consequences: self-expansion and self-extension. The reason for assuming enabling experiences is that, given that SVAs play a central role in linking various devices and IoT home appliances, the interaction between consumers and SVAs is not a one-to-one relationship, but rather part of an interaction experience that comprises a collective that includes various linked objects from the perspective of the assemblage theory (Hoffman & Novak, 2018; Novak & Hoffman, 2019). The assemblage theory is a theoretical perspective that emphasizes what emerges, stabilizes, and destabilizes from the interaction between ontologically equivalent human and nonhuman actors (DeLanda, 2002; Harman, 2002). This study suggests that self-expansion and self-extension, as noted by Hoffman and Novak (2018), are mediating variables for the enabling experience arising from the premise that consumers and SVAs interact within a networked collective. In the following sections, we develop hypotheses on the interaction between consumers and SVAs according to the MEC theory (Figure 1).
Conceptual model and hypotheses
Autonomy is the degree to which an object can operate on its own in a goal-oriented manner without user intervention (Rijsdijk & Hultink, 2009). Autonomy is a core element of smart objects; whether a subject is a smart object depends on whether it has autonomy (Frischknecht, 2021). Without autonomy, objects only respond in a formulaic way according to fixed logical rules and algorithms (De Visser et al., 2018), whereas SVAs can support decision-making on the basis of user preferences and habits through self-learning capabilities (Lucia-Palacios & Pérez-López, 2021) and autonomously sense, think, and monitor when performing tasks, such as activating various functions on other devices or applications (Huang, Kim, & Lennon, 2024).
Interactivity is the degree to which users perceive interaction and communication as two-way, controllable, and responsive to their actions (Lucia-Palacios & Pérez-López, 2021). Like autonomy, interactivity is a key characteristic of smart objects (Lucia-Palacios & Pérez-López, 2023). Two-way communication refers to the exchange of information between users and SVAs. Perceived control refers to the perception that users control their communication with SVAs and foster trust in them (Kang, Shao, & Zhang, 2024). Responsiveness is the perception of the speed of an SVA’s response to a user’s request or question (Song & Zinkhan, 2008). A quick response from their SVAs makes users feel that their opinions are being listened to and respected, which improves their interactive experiences and leads to a positive attitude toward the SVAs.
2. Self-expansionWe discuss self-expansion and self-extension as enabling experiences perceived through interactions with the intelligent attributes of SVAs. Self-expansion is an essential motivation for human beings to strengthen their self-concept by acquiring the resources, perspectives, and identities required to enhance their ability to achieve their goals (Aron & Aron, 1986; Reimann et al., 2012). It can also be described as the degree to which an individual perceives that their self-awareness has been strengthened by acquiring resources, ideas, and identities from another entity (Mao et al., 2019). Self-expansion is not limited to the acquisition of resources, perspectives, and identities from others, but also occurs through interactions with digital spaces such as the Internet (Niu et al., 2023) and metaverses (Nie et al., 2023), as well as digital devices such as mobile phones (Hoffner et al., 2016) and smartwatches (Liu et al., 2022).
The possibility of self-expansion through interaction experiences with IoT products has also been addressed from the perspective of the assemblage theory (Hoffman & Novak, 2018). As the current study targets SVAs that have enhanced autonomy and interactivity by being equipped with AI, they can be expected to promote self-expansion to the same or even greater extent than the targets in previous research.
H1a: Autonomy of SVAs enhances consumers’ self-expansion.
H1b: Interactivity of SVAs enhances consumers’ self-expansion.
3. Self-extensionSelf-extension, another consequence of an enabling experience, refers to an individual’s tendency to engage with an object to define the self (Belk, 1988; Ferraro et al., 2011). When an object plays a role in constructing an individual’s identity, the individual perceives the object as part of the self, thereby promoting self-extension (Park & Kaye, 2019). Self-extension has also been reported for digitized objects and spaces, as well as real objects (Belk, 2013).
While the discussion has focused on self-extension occurring for inanimate objects owned by individuals, with the increasing use of AI in a variety of objects, particularly IoT products, the question of whether smart objects can be entities that achieve self-extension as well as or better than inanimate objects has become a more extensive consideration is needed (Huang & Rust, 2017).
As SVAs learn user preferences and habits autonomously, it is expected that users will increasingly come to see them as entities that understand them better. This will provide an opportunity to obtain a better understanding of the object and build a relationship with it. Knowing an object well is an opportunity to achieve self-expansion (Belk, 1988). Additionally, SVAs, which provide autonomy, enable users to have controllable interactive experiences. The perception that an object is controllable is another opportunity to achieve self-extension (Belk, 1988). On the basis of these findings, SVAs with increased autonomy and interactivity are expected to promote self-extension.
H2a: Autonomy of SVAs enhances consumers’ self-extension.
H2b: Interactivity of SVAs enhances consumers’ self-extension.
4. Consequences and values: Approachability, anthropomorphism, and subjective well-beingSelf-expansion enhances subjective well-being, strengthens a person’s self-concept, and improves mental health (Aron et al., 2001). Therefore, self-expansion has been shown to motivate individuals to maintain and develop relationships with their objects and affect their feelings of closeness with those objects (Aron et al., 2003; Harasymchuk et al., 2021). This has also been confirmed to be the case with inanimate objects, such as smartphones (Hoffner et al., 2016), and people are expected to perceive the same or even greater well-being when living with SVAs, which are inanimate objects with intelligence attributes.
A similar effect can be assumed for self-extension. Self-extension differs from self-expansion in that the former does not strengthen self-concept. However, one’s self-concept is projected onto the object, helping one express oneself and perceive well-being (Park & Kaye, 2019). This has also been confirmed to be the case with inanimate objects, such as smartphones (Ross & Kushlev, 2024). People are expected to perceive the same or even greater well-being when living with SVAs.
H3a: Consumer self-expansion enhances subjective well-being.
H3b: Consumer self-extension enhances subjective well-being.
Here, we focus on the consequences that mediate the relationship between enabling experiences and subjective well-being in terms of perceiving well-being through interactions with SVAs. As SVAs have intelligent attributes, interactions may begin when consumers call out. However, cases in which SVAs actively initiate interactions without interfering with their users’ lives also occur. Additionally, SVAs learn user preferences and habits and sometimes make new suggestions that were not previously made to the user. This SVA behavior can be understood in terms of ambient awareness. Ambient awareness is a concept discussed in the context of human‒computer interaction. For example, fragmented information transmitted via social media, such as SNS, may be meaningless, mundane, and impersonal. However, when such information accumulates over a long period, it can create an awareness of social others. Consumers perceive well-being because ambient awareness does not require conscious effort to acquire information; in the context of social media, it enables the consumer to build social relationships (Krämer et al., 2017).
Ambient awareness is a concept that has developed further owing to research focusing on communication via social media (Levordashka & Utz, 2016). Although differences exist in the interface between social media and SVAs, the structure of how the interaction experience affects the impressions of a target can be considered to be that of ambient awareness, which is formed in the peripheral consciousness by accumulating fragmented information over a long time (in the case of SNS, this would be, for example, posts about the condition of the town where the user currently resides, and in the case of SVAs, this would be information about the current weather, including temperature).
Ambient awareness created through social media increases subjective well-being through feelings of closeness (Krämer et al., 2017). If we view the interaction with SVAs as having the same structure as social media does, we expect users to feel a sense of closeness toward SVAs and perceive well-being in their lives with SVAs.
One antecedent that may lead to a sense of closeness with SVAs is social presence, or the awareness that someone is nearby (Mclean et al., 2021; Mishra et al., 2022). Social presence toward inanimate objects increases the degree of anthropomorphism with which an object is viewed (Schuetzler et al., 2020). Anthropomorphism is the tendency to attribute human characteristics, emotions, and behavior to nonhuman objects, and even if we recognize that an object is inanimate, we can subconsciously perceive it as human (Moon, 2000). Social presence has been reported to be perceived as well-being on social media, such as live streaming, which has increased interaction in recent years (Huang, Yan, & Deng, 2024). This suggests that consumers perceive a greater degree of anthropomorphism when they perceive a sense of social presence, which leads to a sense of closeness, and, therefore, they can be expected to perceive well-being in their lives with SVAs.
Thus, we can conclude that self-expansion and self-extension enable experiences triggered by ambient awareness, which is formed by the intelligent attributes of SVAs, and that the interaction experience makes individuals perceive approachability and anthropomorphism toward SVAs as well as well-being in living with SVAs.
H4a: Consumer self-expansion increases approachability toward SVAs.
H4b: Consumer self-extension increases approachability toward SVAs.
H5a: Consumer self-expansion increases anthropomorphism toward SVAs.
H5b: Consumer self-extension increases anthropomorphism toward SVAs.
H6: As anthropomorphism toward SVAs increases, approachability toward SVAs increases.
H7a: Enhanced approachability toward SVAs increases consumer subjective well-being.
H7b: Enhanced anthropomorphism toward SVAs increases consumer subjective well-being.
This study was conducted online via CrowdWorks in October 2023. The participants were asked whether they had used SVAs (Alexa, Google Assistant, or Siri). We recruited 280 respondents; however, 35 who provided contradictory answers were excluded from the subsequent analysis. Additionally, we excluded respondents who did not speak with their SVAs at least once a day at the time of the survey. Thus, the final number of respondents was 234 (117 females, Mage=38.87 years). The average duration for which respondents had used their SVAs was 2.5 years, and over 50% started using their SVAs in 2022 or later. The average number of conversations with SVAs per day was 4.09. The SVAs used by the respondents were Alexa (73.9%), Google Assistant (19.0%), and Siri (7.1%).
For each measurement item, we used the same items as those used in previous research to ensure scale validity. Nine items measuring autonomy were adopted from Hu et al. (2021) and Kang and Shao (2023). Interactivity was measured via 12 items: four items for two-way communication, four items for perceived control, and four items for responsiveness (Kang & Shao, 2023; Song & Zinkhan, 2008). We used three items each to measure self-expansion and self-extension (Liu et al., 2022). We used the three items adopted from Mishra et al. (2022) for anthropomorphism and the 10-point continuous scale adopted from Levordashka and Utz (2016) for approachability. Finally, we adopted the four items used by Diener et al. (1985) for subjective well-being.
Additionally, this study used two demographic attributes (age and years of experience using SVAs) as control variables. All measurement items except approachability were rated on a 7-point Likert scale ranging from 1 (not at all) to 7 (strongly agree). All the measurement items were in English. However, after being translated into Japanese by a translation agency, they were checked by experts and graduate and undergraduate students for any ambiguity in wording or ease of response and then back-translated (Table 1).
Measurements and Reliability
This study used a structural equation model (CB-SEM: covariance-based structural equation modeling) to verify the hypotheses. First, the reliability of each construct was evaluated via Cronbach’s alpha coefficient and composite reliability (CR). As shown in Table 1, the Cronbach’s alpha and CR coefficients exceeded the criterion value of 0.7 for all the constructs, whereas the average variance extracted (AVE) for the three subconcepts of interactivity (two-way communication, perceived control, and responsiveness) did not slightly exceed the criterion value of 0.5. However, we were generally able to obtain sufficient reliability (Bagozzi & Yi, 1988).
We then assessed the convergent validity of each construct via confirmatory factor analysis with the maximum likelihood estimation method. The goodness-of-fit indices for the model were χ2 [260]=744.438, p<0.001, χ2/df=2.863, SRMR=0.057, IFI=0.884, TLI=0.865 and CFI=0.883; except for one item, the factor loadings for all measurement items exceeded 0.5, and the results for convergent validity were also generally good. As shown in Table 2, the discriminant validity of each construct was good, with the square root of the AVE for each construct being greater than the correlation coefficient between the constructs (Bagozzi, 1981).
Convergent and Discriminant Validity
Note: Diagonal entries are squared root of AVE for each construct. Off-diagonal entries are the correlation coefficients between constructs.
Additionally, as this study collected self-reported data via a within-person cross-sectional design, the possibility of common method bias existed; therefore, we conducted Herman’s one-factor test (MacKenzie et al., 2005). Exploratory factor analysis via the maximum likelihood method extracted seven factors with eigenvalues greater than one, and the proportion of variance of all observed variables explained by these seven factors was 61.0%. The proportion of variance of all the observed variables explained by the first factor with the largest eigenvalue was 37.9%. On this basis, we concluded that the possibility of common method bias was low in this sample. We then tested for common method bias via the marker variable method (Lindell & Whitney, 2001). The variable with the lowest correlation with all the observed variables in this study (gender) was selected and added to the structural equation model for evaluation. The results revealed that the significance of all the predictive paths remained unchanged, indicating that the potential for common method bias was low in this study.
2. Structural equation modelingThe hypotheses were verified via a structural equation model using the maximum likelihood estimation method. The goodness-of-fit indices for the model were χ2 [18]=102.85, p<0.001, χ2/df=5.714, SRMR=0.069, IFI=0.913, TLI=0.828, and CFI=0.912, which were generally good (Figure 2).
Structural equation modeling analysis results
With respect to the effect of SVAs’ intelligent attributes on the interaction experience with consumers, autonomy had a significantly positive effect on self-expansion (β=0.596, p<.001), but no significant effect of interactivity was found (β=0.122, p>.1). Thus, H1a was supported, and H1b was not supported. Similarly, both autonomy (β=0.158, p<.05) and interactivity (β=0.555, p<.001) had significantly positive effects on self-extension. Thus, H2a and H2b were supported.
Self-expansion had no significant effect on approachability toward SVAs (β=−0.082, p>.1), but self-extension had a significant positive effect (β=0.515, p<.001). Thus, H4a was not supported, but H4b was supported. Conversely, while self-expansion had a significantly positive effect on anthropomorphism with respect to SVAs (β=0.491, p<.001), self-extension had a marginally significant effect (β=0.114, p<.1). Thus, H5a was supported, but H5b was not. Additionally, the increase in the degree of anthropomorphism with respect to SVAs increased feelings of closeness (β=0.210, p<.01). Thus, H6 was supported. Therefore, the degree of anthropomorphism toward SVAs increased through the enabling experience of self-expansion, in turn increasing feelings of closeness. Conversely, self-extension did not increase the degree of anthropomorphism toward SVAs but directly increased feelings of closeness.
Finally, this study measured subjective well-being in response to living with SVAs. Self-expansion (β=0.242, p<.001) and self-extension (β=0.473, p<.001) positively impacted subjective well-being directly, thus supporting H3a and H3b, and approachability toward SVAs (β=0.217, p<.001) had an identical positive impact, thus supporting H7a. Conversely, anthropomorphism toward SVAs (β=0.015, p>.1) had no significant effect; thus, H7b was not supported.
This study examined how the two main intelligent attributes of SVAs affect consumer well-being from the perspective of the MEC theory. The results of structural equation modeling via maximum likelihood estimation methods revealed that enabling experiences induced by autonomy and interactivity had different effects on the formation of ambient awareness and subjective well-being.
Autonomy was found to primarily cause consumer self-expansion, and the interaction experience directly enhanced subjective well-being regarding living with SVAs (β=0.144, p<.001). Autonomy was also found to cause consumer self-expansion, which fostered ambient awareness, thus increasing anthropomorphism and approachability toward SVAs and indirectly enhancing subjective well-being with respect to living with them (β=0.016, p<.05).
Autonomy is intelligence that learns a user’s preferences and habits and operates purposefully without user intervention. Therefore, the autonomy of SVAs will lead to the creation of new lifestyles that make consumers’ daily lives more comfortable. Autonomy could be an intelligence attribute that could impact intrinsic human motivation, which is to acquire resources, perspectives, and identities to enhance the capabilities required to achieve goals and strengthen self-concept.
Interactivity was found to cause only consumer self-extension, and the interaction experience likewise directly enhanced subjective well-being toward living with SVAs (β=0.262, p<.001). Interactivity was also found to cause only consumer self-expansion, increasing approachability without anthropomorphism toward SVAs, and indirectly increasing subjective well-being regarding living with SVAs (β=0.062, p<.001).
Interactivity is the perception that interactions and communications with SVAs are two-way, controllable, and responsive to consumer actions. Therefore, if consumers feel comfortable conversing with SVAs and that their views are listened to and respected, this will influence their identity construction. This intelligent attribute incites consumers to perceive SVAs as part of themselves.
With the development of AI, SVAs as well as other kinds of smart objects are becoming increasingly prevalent in our daily lives. In addition to autonomy and interactivity, in future research, we must discuss consumer interaction experiences with all AI properties. Examining negative aspects rather than focusing only on positive aspects is necessary (Vimalkumar et al., 2021). For example, as autonomy develops, the discomfort felt toward SVAs increases; however, this discomfort can be reduced by increasing interactivity (Lucia-Palacios & Pérez-López, 2021; Sidlauskiene, 2022). The results of these previous studies suggest that boundary conditions should be explored further to ensure the acceptance of AI technologies in the future.
The current study focused on the self-expansion and self-extension induced by the interaction experience between consumers and SVAs and how consumers perceive their well-being with respect to living with AI. In this study, we investigated the interaction experience with AI from the consumer’s perspective, but creating an excellent interaction experience for companies in the AI era will also lead to a competitive advantage (Keng et al., 2023). How should we live in harmony with AI, which will continue to develop at an ever-increasing pace? In the era of computers and the Internet, discussions from a human-centered perspective are the norm. However, as we have come to live surrounded by various smart objects, the perspective of the assemblage theory, which states that humans are also a part of this collective, is gaining note (Hoffman & Novak, 2018). Examining the symbiosis between consumers and AI is a research agenda that will continue for some time, and this study should be expanded on to help to progress in this direction.
This research was supported by Grants-in-Aid for Scientific Research (B: 21H00757 and C: 22K01758).
The dataset generated and analyzed in this article is openly available in J-STAGE Data.
Akihiro Nishimoto
A. Nishimoto is Professor of Marketing at the School of Business Administration, Kwansei Gakuin University. He earned his Ph.D. in business administration from Keio University and his MA and BA from Kwansei Gakuin University. His research focuses on the process of market creation and interaction experiences between consumers and AI-enabled smart objects.
https://ror.org/00hhkn466 
