Next Article in Journal
Futures Studies for Design Systems and Social Transformation from Time/Space-Based Perspectives
Previous Article in Journal
Blockchain-Embedded Strategic Options for Satellite Operators: Sell, SataaS, or Dual?
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Systems
Volume 11
Issue 11
10.3390/systems11110551
systems-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Why Do Older Adults Feel Negatively about Artificial Intelligence Products? An Empirical Study Based on the Perspectives of Mismatches

by
Wenjia Hong
1,
Changyong Liang
1,*,
Yiming Ma
2 and
Junhong Zhu
1
1
School of Management, Hefei University of Technology, Hefei 230009, China
2
College of Management Science and Engineering, Anhui University of Finance and Economics, Bengbu 233030, China
*
Author to whom correspondence should be addressed.
Systems 2023, 11(11), 551; https://doi.org/10.3390/systems11110551
Submission received: 12 October 2023 / Revised: 2 November 2023 / Accepted: 9 November 2023 / Published: 15 November 2023
(This article belongs to the Topic Advanced Paradigms, Systems and Enabling Technologies for Product Life Cycle)
Browse Figures
Versions Notes

Abstract

:
Artificial intelligence products (AIPs) for older adults enhance the functions of traditional products and improve the quality of their lives. However, AIPs are not popular among this population, and limited attempts have been made to investigate these users’ negative tendencies regarding AIPs. This study explores the causes of avoidance and exit behaviors toward AIPs among older people from both a functional and socio-emotional mismatch perspective. Data were collected from 1102 older AIP users to verify the research model and hypotheses. The results indicate that perceived control and expectation disconfirmation affect the functional mismatch, while public stigma has the greatest impact on the socio-emotional mismatch. Furthermore, the results highlight a mixed influence of the functional and socio-emotional mismatches on negative behaviors. This study explores older people’s negative tendencies toward AIPs, comprehensively considering the functions of AIPs and the socio-emotions they evoke. Thus, it provides new empirical evidence for the systematic relationship between the functional mismatch and the socio-emotional mismatch and fills the research gap on the influence on the subsequent behaviors of older adults. Additionally, this study sheds light on the specific methods of designing, developing, and promoting AIPs.

1. Introduction

The new generation of information technology has initiated a digital society in which the prominent application of artificial intelligence (AI) benefits humankind. Indeed, the adoption of AI has received considerable attention in almost all sectors, from healthcare to manufacturing [1]; in particular, AI products (AIPs) in elderly care that are equipped with assisted decision-making, automatic response, and intelligent service features are emerging, enhancing the functions of traditional products and improving the quality of the users’ lives [2].
The new generation of conversational artificial intelligence represented by the large language model shows impressive natural language understanding and text generation ability [3], bringing the dawn in the field of voice assistants and robotic companions for older adults. However, there still remain many problems to be solved in the practice of AIPs in the field [3] of assistive technology for older people. In a recent large-scale promotion of AIPs in the Anhui Province, China, many older people shelved or abandoned AIPs after short-term usage, and follow-up surveys revealed attitudes of avoidance or resistance toward AIPs among these users. Furthermore, related studies also confirmed that these tendencies are prevalent among this population with different types of AIPs (e.g., ambient assisted living systems [4], healthcare monitoring products [5], and intelligent monitoring systems [6]). Thus, exploring the reasons for these negative tendencies of AIPs, rather than the reasons for adoption, could be more important. This will promote the deep integration of AI with human lives and solve social problems brought by AI.
Previous research has shown that the mismatch between the requirements of older people for the products they use and the AIPs’ actual functions, along with the mismatch between their expectations and the socio-emotions the AIPs evoke, explain the failure of the AIP promotion [7,8]. For one thing, a decline in autonomy and a rise in expectations of independence with increasing age become the root of many conflicting events in the lives of older adults [9]. The assisted decision-making and intelligent service functions that AIPs possess, to a certain extent, reduce the opportunities for older people to make their own decisions and deviate from their actual needs, which has resulted in a negative attitude toward AIPs. Additionally, AIPs offer social significance to users while meeting their functional needs [10]. However, a mismatch can occur between users’ expectations of AIP and the social evaluation they receive from using it, influenced by algorithmic bias, ageism, and stigma [11]. Accordingly, it is necessary to explore the behaviors of the avoidance and exit of AIPs among older people by considering both the functional mismatch and the socio-emotional mismatch.
Existing studies have mostly focused on a single dimension of products, such as the functions of products or the socio-emotions they evoke. For example, a research paradigm based on product functional design discusses participatory design [12], inclusive design, and accessible design for older adults by considering human–computer interaction [13]. However, these studies fail to spotlight the functional mismatch caused by the perceived intrusiveness and control of AIPs, along with the impact of the mismatch. In addition, several studies have been conducted from the socio-emotional perspective, for example, research on the mechanisms by which public stigma affects the resistance to AIPs of older people [14], research on how to dispel negative stereotypes of older people being unhealthy and technologically illiterate [7], and on the reasons for the negative feelings that arise (e.g., self-stigma and inferiority) when older people use AIPs [15]. While these studies focus on the causes of the socio-emotional mismatch, they lack empirical evidence connecting this mismatch to subsequent behavior.
In conclusion, no systematic research to date has explored the causes of the mismatches between AIPs and older people, and research gaps still exist on the impact of the mismatches on subsequent behavior by older users. Thus, this paper poses the following research questions to systematically understand the causes of functional mismatch and socio-emotional mismatch, along with the mechanisms that affect the negative AIP usage behavior of older people.
RQ1: What factors influence the functional mismatch and socio-emotional mismatch in the use of AIPs among older people?
RQ2: What are the differences in the mechanisms by which this functional mismatch and socio-emotional mismatch influence the negative use behavior among older people?
To address these issues, we construct a research model based on the cognition–affect–conation (CAC) pattern to investigate the causes of the avoidance and exit behaviors toward AIPs among older adults from the perspectives of the functional mismatch and socio-emotional mismatch in order to support future deep application of AIPs. In the next section, this paper reviews the existing literature and presents the research hypotheses. The third section describes the research methodology. The fourth section explains the data analysis and presents the results. Discussion, implications, and limitations are reported in section five. Finally, the last section presents conclusions.

2. Literature Review and Hypothesis Development

2.1. The Cognition–Affect–Conation Pattern

The CAC pattern describes the effect of external stimuli on an individual’s emotions and intentions [16]. Cognition is the intellectual, spiritual, or rational state of an individual. Affect is the responses and preferences of the individual to external stimuli on the basis of cognition. Conation is a behavioral tendency to make comprehensive decisions based on cognition and affect. The essence of the CAC framework is the process of “information processing → preference formation → behavioral tendencies (actual behavior)” [17]. The CAC framework illustrates the direct influence of cognition on affective outcomes (e.g., attitudes or satisfaction) to motivate individual behavior [16].
The CAC framework, which is often used to study consumer behavior, provides a multi-level perspective to understand the mechanisms by which individuals’ cognition about products is translated into actions through effect and enables examination of the impact of technology in the decision-making process of consumers [18]. Furthermore, the CAC framework is already being used as the foundation of other theories that can effectively explain the consumer adoption and usage behavior of new technologies [19]. According to previous studies, the CAC framework is well suited to explain the impact of new technologies on individual behavioral intentions, in both the functional and socio-emotional perspectives, for example, students’ continuance intention on using mobile social network sites [20] or users’ continued use and purchase intentions toward augmented reality technology [19]. At the same time, the CAC framework is able to explain the mechanism that generates negative usage behaviors of technologies and products, such as the avoidance of social media [16] or passive usage [21]. Therefore, using the CAC framework to study older people’s avoidance and exit behaviors towards AIPs is appropriate.

2.2. Negative Effects of AIPs for Older People

AIPs are a new generation of information technology products that can interact with the environment and are designed to mimic human intelligence [22]. AIPs for older people, such as ambient assisted living systems and healthcare monitoring devices, provide potential solutions and new opportunities to reduce the burden of elderly care and improve the quality of their lives [23]. However, AIPs equipped with assisted decision-making and intelligent services often result in negative emotions because they do not match the actual needs of older people. In addition, the designers of AIPs have difficulty accurately analyzing and understanding the emotions of older people due to a lack of training data from this population, leading to negative tendencies toward AIPs among older people. Specifically, it is not always beneficial for older people to be overloaded with advice and assistance [24]. The gradual loss of independent living skills with increasing age makes older people more eager to make decisions on their own, so the frequent advice from the AIPs is often seen as an interruption or intrusion [25]. Furthermore, AIPs lead to negative effects on older people’s self-esteem, perceived control, and psychological cognition [26,27]. Therefore, we review the current research to further identify and confirm the negative effects of AIPs on older people (Table 1).
As Table 1 shows, the negative effects of AIPs are diverse, including technical, social, and psychological aspects. AIPs have functional properties, indicating that some negative effects can be categorized as being in the functional dimension [40]. In addition, the social attributes of AIPs, such as the expectations of individuals and the perceptions of others, show a more social, psychological, and emotional dimension than other types of products [41]. Therefore, we divide the negative impact of AIPs into two aspects: a functional mismatch and a socio-emotional mismatch.

2.3. Functional Mismatch of AIPs

User experience, the satisfaction of requirements, and expectation confirmation are the basic principles of modern product design [42]. For older people, the core design principle of age-friendly products is to meet their relevant needs based on their physiological and psychological characteristics [43]. In the “pre-smart” era, safety, usability, and attractiveness were the key factors considered by older people when choosing products [44]. However, the design of AIPs focuses more on having functions, such as assisted decision-making and intelligent services, all accomplished using big data and intelligent algorithms [45]. Although current AIPs for older people meet the design principle of “satisfaction of requirements”, the current design philosophy may be flawed; as a result, the products deviate from the expectations and experiences of older people [46].
The physical deterioration caused by aging can make daily life progressively more difficult for older people. AIPs for older adults were originally designed to help them overcome these difficulties and maintain their autonomy and independence. However, these AIPs often trigger negative emotions by not matching the actual needs of the users [47]. First, the perceived intrusiveness of the AIPs becomes a social issue because the AIPs tend to acquire information and preferences of individuals to provide personalized services and achieve targeted functions [48]. Older people are more concerned than others about this invasion of personal privacy and security enough that they resist using AIPs [32]. Additionally, privacy concerns make it difficult for older people to enjoy the personalized, high-quality services offered, and ultimately, older users have more negative emotions about the functions of the AIPs. Second, older people expect to be assisted by AIPs. However, when the AIP offers too many reminders and suggestions, the user feels increasingly controlled by it; this phenomenon is more obvious for older people who are eager to live independently, which leads to a series of negative comments on the functions of the AIPs by older people [33]. In addition, older people tend to misunderstand or have excessive expectations about the actual functions of AIPs due to their lack of information technology literacy [49]. However, AIPs have difficulty meeting these expectations, leaving a mismatch between the AIP functionality and the user requirements [50]. Therefore, we propose the following hypotheses about AIPs and older users.
H1. 
Perceived intrusiveness can significantly affect the functional mismatch.
H2. 
Perceived control can significantly affect the functional mismatch.
H3. 
Expectation disconfirmation can significantly affect the functional mismatch.

2.4. Socio-Emotional Mismatch of AIPs

Except for functional attributes, the AIPs possess many social attributes, such as a user image, social evaluation, and memories related to products [51]. Thus, the socio-emotional mismatch of individuals toward products is often determined by social evaluation and individual responses that AIPs evoke [52]. For older people, AIPs often amplify negative feelings that are inconsistent with the social identification expected from using them, influenced by digital ageism and algorithmic bias. For one thing, the use of age-friendly products is in itself a reflection of public stigma (the prejudice and discrimination endorsed by the general population that affects a person) [4,53]. Although people are sympathetic to older adults, implicit discrimination is unavoidable. In particular, the AIPs are often connected to the relatives of the users or their communities or hospitals. Not only does this connection make more people aware of the aging and weakness of the user, but it also reinforces the stereotypes and stigma. Therefore, older people feel more negatively about the social significance of the AIPs based on social feedback and evaluations. For another, the internalization of public stigma among older people increases, namely, self-stigma (the harm that occurs when the person internalizes the prejudice) [53], influenced by ageism and implicit discrimination based on big data [15]. This increase in self-stigma affects an older user’s capability of independent living and social participation. And the pressure brought by the stigma is likely to form a vicious circle that makes it difficult for older adults to reconstruct their positive self-image and reduces their social adaptability, leading to a socio-emotional mismatch. Additionally, in the information and communications technology field, technology anxiety has an equal effect on the socio-emotional mismatch [54]. Since older people cannot understand the logic of AIPs and the operation mechanism of digital society reflected by this logic, they are anxious about the social isolation implied by the use of AIPs, which leads to further socio-emotional mismatch. Based on all these factors, we propose the following hypotheses.
H4. 
Self-stigma can significantly affect the socio-emotional mismatch.
H5. 
Public stigma can significantly affect the socio-emotional mismatch.
H6. 
Technology anxiety can significantly affect the socio-emotional mismatch.

2.5. Avoidance and Exit Behavior

According to previous studies, the primary reason for avoidance or exit behavior is a mismatch between products and the requirements of consumers [40], including a functional mismatch and a socio-emotional mismatch [55]. In the digital society, older adults are using AIPs actively or passively, but this does not mean that older people accept or prefer to use the AIPs [56]. Specifically, according to Ho’s research, despite the good applied value of AIPs in elderly care, biases based on functional and socio-emotional mismatch could lead to avoidance behavior of AIPs among older people [57]. In addition, Ploug and Holm’s research showed that AI bias and discrimination in disease diagnosis and treatment, as well as patient preference for face-to-face treatment, can significantly increase exit tendency by older people [57]. Dwivedi et al. pointed out that functional mismatch and socio-emotional mismatch remained major challenges that prevent the promotion of AIPs among older people [6]. In summary, we argue that a functional mismatch and socio-emotional mismatch in the use of AIPs can lead to avoidance and exit behaviors among older people. Therefore, we hypothesize as follows.
H7. 
Functional mismatch can significantly affect the avoidance behavior of older people.
H8. 
Functional mismatch can significantly affect the exit behavior of older people.
H9. 
Socio-emotional mismatch can significantly affect the avoidance behavior of older people.
H10. 
Socio-emotional mismatch can significantly affect the exit behavior of older people.
We examined older adults’ avoidance and exit behaviors toward AIPs from the perspectives of two mismatches based on the CAC framework. As shown in Figure 1, we propose that the perceived intrusiveness, perceived control, and expectation disconfirmation of AIPs affect the functional mismatch; self-stigma, public stigma, and technology anxiety affect the socio-emotional mismatch. The functional mismatch and socio-emotional mismatch can lead to the avoidance and exit behaviors of older people toward AIPs.

3. Materials and Methods

3.1. Survey Instruments

All scales were based on previous studies and modified to fit the current research context. All measurement items adopted a seven-point Likert scale where 1 represents negative (strongly disagree or strongly agree), and 7 represents positive (strongly agree or strongly disagree), with the exception of the control variables. The measurement items of perceived control were adapted from Lachman and Weaver [58]; the items of perceived intrusiveness were adapted from Xu et al. [59]; the items of public stigma and self-stigma were adapted from Vogel [60]; the items of technology anxiety were adapted from the results of Jeng, Pai, and Yeh [61]; the items of expectation disconfirmation were adapted from Lin, Wu, and Tsai [62]; the items of functional mismatch and socio-emotional mismatch were adapted from Chen’s research [15]; and the measures of avoidance and exit behaviors were adapted from Ogbanufe and Gerhart [63]. Table 2 shows the variables and indicators.

3.2. Sample and Data Collection

This study collected firsthand data using a questionnaire survey to verify the conceptual model. Potential respondents were individuals over 60 years old who had used at least one kind of AIP. Given that some measurement items were not available in Chinese, the English items were first translated into Chinese by one author who was proficient in both languages. The items were then tested by two experts in AI and behavioral sciences and seven PhDs in related areas. To ensure the quality of the questionnaire, 50 copies of a pre-survey questionnaire were distributed offline. All 50 were collected. Some respondents were interviewed, and the questionnaire was revised according to their suggestions.
The formal questionnaire was collected in March–April 2022 with the assistance of the Civil Affairs Bureau of Hefei (Hefei, Anhui Province, China). With the help of community workers, we distributed questionnaires to older people who had participated in a previous, large-scale promotion of AIPs. Since the current average retirement age in China is 55 years old, when people retire, their behavior and psychology have become de facto elderly people. In the relevant research on the elderly and artificial intelligence, a large number of scholars also set the age of the research object as 55 years old [35,64,65]. In the previous promotion activities, the target of distribution was also 55 years old. Therefore, the study set the starting age of respondents as 55 years old.
Relevant training was given to all volunteers before the distribution. Each questionnaire took about 30 min to complete; for those participants who had difficulty reading and writing, the questionnaires were completed using an oral question-and-answer format with the assistance of volunteers. In addition, demographic information such as the respondent’s age, gender, and educational level were required in the questionnaire. A total of 1574 questionnaires were collected, and after eliminating the invalid questionnaires, 1102 valid questionnaires were received, with an effective rate of 70.01%. The statistical characteristics of the sample are shown in Table 3.

4. Data Analysis and Results

We used partial least squares (PLS) methods for data analysis in an exploratory study to simultaneously assess the reliability and validity of construct measures and to estimate the relationship between constructs. In addition, PLS imposes minimal restrictions on sample size and residual distribution. We followed a two-step approach when examining the measurement and structural models.

4.1. Measurement Model Testing

4.1.1. Common Method Biases and Multicollinearity

The variance inflation factor (VIF) was measured. The VIF can measure the severity of collinearity in multiple linear regression models. Although a VIF value of 3.3 or 5 is the best under ideal conditions, a large number of literatures also believe that in the study of PLS-SEM, a VIF less than 10 is sufficient [66,67]. According to Shrestha (2020), when the VIF value is between 5 and 10, it is still acceptable despite the challenging value [68]. In this study, the maximum value of the VIF was 7.295, which is less than the threshold value of ten, indicating that the model does not have a multicollinearity problem.
To ensure the fit validity of the model, the standardized residual root mean square (SRMR), unweighted least square difference (d_ULS), and geodesic difference (d_G) of the model were measured to determine how well the model fitted the data. SRMR should be less than 0.08, and SRMR, dULS, and dG should be less than 95% of the bootstrap difference. As shown in Table 4, the SRMR value is 0.034, which is less than the threshold value of 0.08. The values of SRMR, d_ULS, and d_G are all less than 95% bootstrap differences. Overall, the analysis showed that, with a 5% probability, the measurements were sufficient for empirical analysis.
To ensure that there is no common method bias in the samples, Harman’s one-factor test was used for measurement in this study [69]. In Harman’s one-factor test, when the ratio of the largest single factor to the variance is less than 50%, it is generally considered that common method bias is not supported [70]. SPSS22 was used for analysis, and the results showed that the maximum factor accounted for 27.785% (less than 50%), and the total factor (nine factors) accounted for 72.601%. In addition, this paper uses the marker variable method to test the common method biases [71]. The results show that each marker variable has no significant influence on the model variables. Therefore, the possibility of common method deviation in this study is small.

4.1.2. Reliability and Validity

To validate the measurement model, we assessed the reliability of the construct and two types of validity (convergent validity and discriminant validity). Table 5 shows the factor loading, combined reliability (CR), Cronbach’s alpha, and average variance extracted (AVE) coefficient of the model. In this study, the minimum value of factor loading is 0.779, and the maximum is 0.96, both of which are greater than 0.7 [72]. As an effective indicator of the internal reliability of each dimension of the model, CR plays an important role in evaluating the model. In this study, the minimum value of CR is 0.897, and the maximum value is 0.972, which is greater than the recommended value of 0.7 [72]. Cronbach’s alpha is an important measure of the internal validity of the model, and most of the literature suggests that the value needs to be greater than 0.7 [72]. The range of values in this study is 0.827–0.962, which is greater than the threshold value of 0.7, indicating good internal consistency of the questionnaire. The AVE took values ranging from 0.743 to 0.898, which are greater than 0.5, indicating that the observed items explain much more variance than the error term and that the validity of the model aggregation is relatively high [72].
The results in Table 6 show that the square root of the AVE for each construct is greater than the correlation involving that construct, which confirms the discriminant validity. According to Hair, the heterogenic–parthenosexual correlation ratio (HTMT) is another method for evaluating discriminant validity [72]. The HTMT value is less than 0.85, indicating that the validity of the discrimination has been determined. In Table 7, all HTMT values are less than 0.85. In summary, the model has good reliability and validity.

4.2. Structural Model

SmartPLS 3.32 was used to test the structural model. Bootstrapping was used, and the maximum number of iterations was 5000. The specific results are shown in Figure 2 and Table 8.
Finally, we conducted a control variable test. The t-test results show that demographic characteristics (age, gender, education, and type of AI used) have no significant impact on the results of the analysis. The t-test results show that age, gender, education, and type of AI used have no significant impact on this research. The results show that demographic characteristics have no significant influence on the results of the analysis.
In addition, to evaluate whether the proposed model has predictive ability, we evaluated whether Q2 is greater than 0; a positive Q2 value shows that the predictive error of the PLS results is less than that of using only the mean value [73]. In this model, Q2 ranges from 0.204 to 0.550, all of which are greater than 0, indicating that the model has good predictability.

5. Discussion

5.1. Discussion

This study explored the causes of negative behaviors of older people toward AIPs from the perspectives of a socio-emotional mismatch and functional mismatch. We constructed a structural equation model, tested our hypotheses, and drew the following conclusions:
(1) The functional mismatch is affected by expectation disconfirmation (H3, β = 0.275, p < 0.001) and perceived control (H2, β = 0.363, p < 0.001); however, perceived intrusiveness does not share any relationship with it (H1, β = 0.043, p = 0.303). As suggested in previous research, the excessive reminders and suggestions of AIPs, as well as unrealistic advertisements by AIP sellers, result in the variables of perceived control and expectation disconfirmation, aggravating the functional mismatch [33]; this mismatch reflects older people’s yearning for independence. The invalidity of H1 highlights the choice AIP users must make between privacy and functional needs. This is because AIPs must acquire the personal information and preferences of users to provide personalized service and realize their target functions [74]. That is, the more private information obtained by the AIPs, the more likely the AIPs will be able to meet the requirements of the users. The initial intention of older people to use AIPs affects their choice: they are more inclined to concede their privacy when the expected functions of the AIPs match their actual needs so as to acquire better effects. Essen’s study reached the same conclusion, with older adults viewing a home care monitoring device as freeing and protecting their privacy, as the device enabled them to continue living in their own homes rather than moving to a nursing home [75]. This valuable finding shows that older people are willing to concede some privacy to acquire the expected functions of AIPs, reflecting older people’s desire to satisfy their life needs and further proving H2 and H3;
(2) The socio-emotional mismatch is affected by public stigma, self-stigma, and technology anxiety. However, in terms of specific impact, public stigma has the greatest impact (H4, β = 0.549, p < 0.001); second is technology anxiety (H6, β = 0.162, p = 0.001); and third is self-stigma (H5, β = 0.154, p < 0.001). We believe that the reason for H4 is that older people often experience social prejudices and stereotypes in the process of using AIPs, such as that older adults are weak, unhealthy, and technologically illiterate. In addition, the over-praise of older people when they do use AIPs is a potential stigmatization and likewise leads to negative perceptions of the social significance of AIPs among the users [76]. Older people often feel further stigmatized in the digital society since the level of digital social participation increases when they use AIPs. Therefore, public stigma has the greatest impact on the socio-emotional mismatch. In addition, older people are more likely to internalize the public stigma, causing fundamental damage to their self-image in regard to the digital age and making it difficult for them to integrate into digital society from a psychological level [15]. This valuable finding shows that social opinions and potential negative evaluations do serious harm to older people in the process of using AIPs;
(3) For older people, a significant difference exists in the impacts of the functional and socio-emotional mismatches on negative behaviors toward AIPs. On the one hand, both the functional mismatch (H7, β = 0.363, p < 0.001) and socio-emotional mismatch (H9, β = 0.308, p < 0.001) lead to avoidance behavior. This is consistent with previous research showing that older people have avoidance intentions toward AIPs, driven by their efforts to reduce their cognitive load, avoid the threat of stigma, and regulate their emotions [16], and helps to explain this negative behavior among older adults. On the other hand, while the functional mismatch significantly affects the exit behavior of older people (H8, β = 0.63, p < 0.001), the socio-emotional mismatch does not share any relationship with the exit behavior (H10, β = 0.069, p = 0.2). We believe that the reason for this is that functional satisfaction and perceived usefulness are the main requirements for products for older people [77]. Therefore, if a functional mismatch occurs with AIPs, older people turn to an alternative, useful product and abandon the AIP. The same is not true of a socio-emotional mismatch. Although a socio-emotional mismatch may cause a series of negative effects, such as depression, anxiety, and injury to the user’s self-esteem [15], older users are conservative and often reluctant to give up a product once they become expert at using it. Accordingly, it is unlikely that older people will abandon AIPs completely when they can use the AIPs effectively and the AIP functions satisfy user requirements. This valuable finding suggests that designers should focus on the suitability of AIP functions to the needs of older adults.

5.2. Implications for Research

The present study has outlined three contributions to the theory and the literature.
(1) In this study, the mismatch between products and requirements/expectations is divided into a functional mismatch and a socio-emotional mismatch, and the mechanism of each mismatch’s influence on negative behavior toward AIPs is systematically analyzed. The results show significant differences between the two kinds of mismatch in the avoidance and exit behaviors of older adults. Our study comprehensively considers the functions of AIPs and the socio-emotions they evoke in order to solve the problem of adoption by older adults, which provides new empirical evidence for the systematic relationship between the functional mismatch and the socio-emotional mismatch. It fills the gap in research on the influence of these two factors on the behavior toward AIPs of older people. In addition, the application of the CAC framework to study the interaction between older people and AIPs fully explains why AIPs are not popular among older people and proves the value of applying the CAC framework in the field of information system behavior and artificial intelligence;
(2) This study analyses the causes of the functional mismatch in AIPs, and the results show that perceived control and expectation disconfirmation have a significant impact on the functional mismatch, which provides new empirical evidence for the design theory of AIPs. However, perceived intrusiveness does not share any relationship with the functional mismatch. This finding indicates that seeking an effective balance point between privacy concerns and intelligent services for older adults is an issue for future research and provides new insights for theories of privacy protection among older users. On the flip side, this study explores the reasons for the socio-emotional mismatch, and the results show that public stigma is the most important source. This finding contributes to the development of theories about stigma in the digital society, increases the ways of bridging the digital divide, enriches the research on the psychological integration of older adults in digital society, and improves the theory of social integration;
(3) The last valuable theoretical contribution of this study is about “Technologies in Service of Humanity”. According to Kotler, digital technology, especially AI technology, should serve human beings [78]. In this study, our findings reveal, from a theoretical perspective, why older groups are reluctant to use the new generation of information technology products based on AI technology. Because AIPs have both emotional and functional properties, people’s perception of them is more complex than other types of products. The results of this study show the important impact of functionality and social emotion of AIPs on the elderly, providing a valuable research perspective from the aspects of social opinion, product design, user behavior, etc., and providing new empirical evidence for the expansion of relevant theoretical research in a friendly digital society.

5.3. Practical Implications

This study has three practical contributions.
(1) The results show that both the functional mismatch and socio-emotional mismatch have a significant impact on negative behavior toward AIPs among older people and confirm that the functional mismatch is a key factor in older adults’ exit behavior. This shows that current AIPs are not able to achieve the goal of “Technologies in Service of Humanity”. Based on this, we propose that designers should focus on the functional development of AIPs and then design products that are close to users’ actual requirements. In addition, the large language model should be applied to improve existing AIPs, especially the friendliness and efficiency of human–computer interaction. This will help AIPs better adapt to the requirements of older people and help older people bridge the digital divide and integrate into digital society;
(2) The findings reflect the significant influence of perceived control on the functional mismatch, which conversely demonstrates the desire of older people for independent living and the importance to older people of maintaining their independence and autonomy with the help of AIPs. Therefore, families and organizations need to safeguard the independence of older people. Families are encouraged to solve the problem of AIP accessibility and support older people in using AIPs from informational, emotional, and other aspects. The government should provide policy and financial support to improve the digital literacy of older adults. In addition, nongovernmental organizations such as senior universities and community centers need to play an active role in guiding and communicating with older people in the process of using AIPs to better promote the use of AIPs by older people;
(3) The results emphasize that public stigma is the most important factor of the socio-emotional mismatch. Older people’s lives are losing significance in the digital age due to stigma, marginalization, and a lack of belonging and security. It is difficult for AIPs to satisfy the social and self-identification needs of older people due to algorithmic bias and age discrimination. To alleviate the negative impact of stigma on older people, designers of AIPs can help older adults better use them through embedding large language models to improve AI algorithms, improve older people’s desire to explore the digital society, maintain the positive image of older adults consciously and change the prejudice and discrimination against them. The mainstream media can shape a positive image of older people and bridge the inter-generational digital gap.

5.4. Limitations and Future Research Directions

The study has three limitations. First, data in this paper came from AIP users in Hefei, Anhui Province, China. Due to regional, cultural, and economic differences, the results of this study may not extend to other countries with different cultural systems. Future studies should further examine the influence of cultural differences on the relationship between different older groups (country, ethnicity, and region) and AIPs. Second, the study did not consider the role of family members in helping older people. As an important influencing factor for the elderly to use AIPs, family members and experts can help the elderly better accept and use AIPs [79,80]. Therefore, in future studies, we will further explore the influence of family members on the use behavior of AIPs in the elderly. Finally, due to the heterogeneity of different elderly groups’ needs, cognitive abilities, and other characteristics, the research results in this paper cannot effectively reflect such differences. Therefore, in future studies, we will consider this difference, collect data on elderly people with different characteristics, and use multi-group comparative analysis for further research.

6. Conclusions

Although AI and its applications have penetrated every corner of society, most AIPs face the reality of being shelved or abandoned after short-term usage by older people. Existing research has shown that the mismatch between the requirements of older people for the products they use and the AIPs’ actual functions, along with the mismatch between their expectations and the socio-emotions the AIPs evoke, explain this phenomenon. Hence, this study explored the causes of avoidance and exit behaviors from both a functional mismatch and socio-emotional mismatch perspective. The results have shown that (1) the functional mismatch is influenced by expectation disconfirmation and perceived control; (2) public stigma has the greatest impact on the socio-emotional mismatch; and (3) both the functional mismatch and socio-emotional mismatch lead to avoidance behavior, while only the functional mismatch affects the exit behavior of older adults. Based on these findings, we propose that all organizations should work together to help older people maintain their independence and mitigate the negative effects of stigma. In addition, our study offers theoretical implications of the CAC framework, product design, privacy protection, and social integration, as well as practical implications for designing, developing, and promoting AIPs. These implications can help improve the adaptation of AIPs to older users and promote the active use of AIPs by older people for integration into digital society. These implications can help improve the adaptation of AIPs to older users and promote the active use of AIPs by older people for integration into a digital society, contributing to “Technologies in Service of Humanity”.

Author Contributions

Conceptualization, W.H., Y.M. and C.L.; methodology, W.H. and Y.M.; software, W.H.; validation, W.H., Y.M. and C.L.; investigation, W.H.; resources, W.H. and J.Z.; data curation, W.H.; writing—original draft preparation, W.H. and Y.M.; writing—review and editing, W.H. and Y.M.; visualization, Y.M.; supervision, C.L. and J.Z.; project administration, C.L.; funding acquisition, C.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Office for Philosophy and Social Sciences of China, grant number 21AZD116, and the APC was funded by the National Office for Philosophy and Social Sciences of China, grant number 21AZD116.

Data Availability Statement

The datasets generated and analyzed during the current study are not publicly available due to the fact that they constitute an excerpt of research in progress but are available from the authors upon reasonable request.

Acknowledgments

The authors would like to extend their warm appreciation to the anonymous reviewers and Editor for their valuable comments to improve the quality of this work.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Haibe-Kains, B.; Adam, G.A.; Hosny, A.; Khodakarami, F.; Waldron, L.; Wang, B.; McIntosh, C.; Goldenberg, A.; Kundaje, A.; Greene, C.S. Transparency and reproducibility in artificial intelligence. Nature 2020, 586, E14–E16. [Google Scholar] [CrossRef] [PubMed]
  2. Sohn, K.; Kwon, O. Technology acceptance theories and factors influencing artificial Intelligence-based intelligent products. Telemat. Inform. 2020, 47, 101324. [Google Scholar] [CrossRef]
  3. van Dis, E.A.; Bollen, J.; Zuidema, W.; van Rooij, R.; Bockting, C.L. ChatGPT: Five priorities for research. Nature 2023, 614, 224–226. [Google Scholar] [CrossRef] [PubMed]
  4. Yusif, S.; Soar, J.; Hafeez-Baig, A. Older people, assistive technologies, and the barriers to adoption: A systematic review. Int. J. Med. Inform. 2016, 94, 112–116. [Google Scholar] [CrossRef]
  5. Rubeis, G. The disruptive power of artificial intelligence. Ethical aspects of gerontechnology in elderly care. Arch. Gerontol. Geriatr. 2020, 91, 104186. [Google Scholar] [CrossRef]
  6. Dwivedi, Y.K.; Hughes, L.; Ismagilova, E.; Aarts, G.; Coombs, C.; Crick, T.; Duan, Y.; Dwivedi, R.; Edwards, J.; Eirug, A. Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy. Int. J. Inf. Manag. 2021, 57, 101994. [Google Scholar] [CrossRef]
  7. Chu, C.H.; Nyrup, R.; Leslie, K.; Shi, J.; Bianchi, A.; Lyn, A.; McNicholl, M.; Khan, S.; Rahimi, S.; Grenier, A. Digital Ageism: Challenges and Opportunities in Artificial Intelligence for Older Adults. Gerontologist 2021, 62, 947–955. [Google Scholar] [CrossRef]
  8. Chen, L.-K. Gerontechnology and artificial intelligence: Better care for older people. Arch. Gerontol. Geriatr. 2020, 91, 104252. [Google Scholar] [CrossRef]
  9. De Angeli, A.; Jovanović, M.; McNeill, A.; Coventry, L. Desires for active ageing technology. Int. J. Hum.-Comput. Stud. 2020, 138, 102412. [Google Scholar] [CrossRef]
  10. Song, X.; Xu, B.; Zhao, Z. Can People Experience Romantic Love for Artificial Intelligence? An Empirical Study of Intelligent Assistants. Inf. Manag. 2022, 59, 103595. [Google Scholar] [CrossRef]
  11. Berridge, C.; Grigorovich, A. Algorithmic harms and digital ageism in the use of surveillance technologies in nursing homes. Front. Sociol. 2022, 7, 957246. [Google Scholar] [CrossRef] [PubMed]
  12. Merkel, S.; Kucharski, A. Participatory design in gerontechnology: A systematic literature review. Gerontologist 2019, 59, 16–25. [Google Scholar] [CrossRef] [PubMed]
  13. Stephanidis, C.; Salvendy, G.; Antona, M.; Chen, J.Y.; Dong, J.; Duffy, V.G.; Fang, X.; Fidopiastis, C.; Fragomeni, G.; Fu, L.P. Seven HCI grand challenges. Int. J. Hum.-Comput. Interact. 2019, 35, 1229–1269. [Google Scholar] [CrossRef]
  14. Park, E.H.; Werder, K.; Cao, L.; Ramesh, B. Why do Family Members Reject AI in Health Care? Competing Effects of Emotions. J. Manag. Inf. Syst. 2022, 39, 765–792. [Google Scholar] [CrossRef]
  15. Chen, K. Why do older people love and hate assistive technology?—An emotional experience perspective. Ergonomics 2020, 63, 1463–1474. [Google Scholar] [CrossRef]
  16. Dai, B.; Ali, A.; Wang, H. Exploring information avoidance intention of social media users: A cognition–affect–conation perspective. Internet Res. 2020, 30, 1455–11478. [Google Scholar] [CrossRef]
  17. Lee, L.W.; Tang, Y.; Yip, L.S.; Sharma, P. Managing customer relationships in the emerging markets–guanxi as a driver of Chinese customer loyalty. J. Bus. Res. 2018, 86, 356–365. [Google Scholar] [CrossRef]
  18. Tsiotsou, R.H. Sport team loyalty: Integrating relationship marketing and a hierarchy of effects. J. Serv. Mark. 2013, 27, 458–471. [Google Scholar] [CrossRef]
  19. Qin, H.; Osatuyi, B.; Xu, L. How mobile augmented reality applications affect continuous use and purchase intentions: A cognition-affect-conation perspective. J. Retail. Consum. Serv. 2021, 63, 102680. [Google Scholar] [CrossRef]
  20. Wang, R.-B.; Du, C.-T. Mobile Social Network Sites as innovative pedagogical tools: Factors and mechanism affecting students’ continuance intention on use. J. Comput. Educ. 2014, 1, 353–370. [Google Scholar] [CrossRef]
  21. Li, J.; Guo, F.; Qu, Q.-X.; Hao, D. How Does Perceived Overload in Mobile Social Media Influence Users’ Passive Usage Intentions? Considering the Mediating Roles of Privacy Concerns and Social Media Fatigue. Int. J. Hum.-Comput. Interact. 2021, 38, 983–992. [Google Scholar] [CrossRef]
  22. Glikson, E.; Woolley, A.W. Human Trust in Artificial Intelligence: Review of Empirical Research. Acad. Manag. Ann. 2020, 14, 627–660. [Google Scholar] [CrossRef]
  23. Qian, K.; Zhang, Z.; Yamamoto, Y.; Schuller, B.W. Artificial intelligence internet of things for the elderly: From assisted living to health-care monitoring. IEEE Signal Process. Mag. 2021, 38, 78–88. [Google Scholar] [CrossRef]
  24. Van Weert, J.; Van Munster, B.C.; Sanders, R.; Spijker, R.; Hooft, L.; Jansen, J. Decision aids to help older people make health decisions: A systematic review and meta-analysis. BMC Med. Inform. Decis. Mak. 2016, 16, 45. [Google Scholar] [CrossRef]
  25. Siau, K.; Wang, W. Artificial intelligence (AI) ethics: Ethics of AI and ethical AI. J. Database Manag. 2020, 31, 74–87. [Google Scholar] [CrossRef]
  26. Rosales, A.; Fernández-Ardèvol, M. Structural ageism in big data approaches. Nord. Rev. 2019, 40, 51–64. [Google Scholar] [CrossRef]
  27. Zhu, J.; Shi, K.; Yang, C.; Niu, Y.; Zeng, Y.; Zhang, N.; Liu, T.; Chu, C.H. Ethical issues of smart home-based elderly care: A scoping review. J. Nurs. Manag. 2021, 30, 3686–3699. [Google Scholar] [CrossRef]
  28. Damant, J.; Knapp, M.; Freddolino, P.; Lombard, D. Effects of digital engagement on the quality of life of older people. Health Soc. Care Community 2017, 25, 1679–1703. [Google Scholar] [CrossRef]
  29. Laplante, P.; Milojicic, D.; Serebryakov, S.; Bennett, D. Artificial intelligence and critical systems: From hype to reality. Computer 2020, 53, 45–52. [Google Scholar] [CrossRef]
  30. Vollmer Dahlke, D.; Ory, M.G. Emerging opportunities and challenges in optimal aging with virtual personal assistants. Public Policy Aging Rep. 2017, 27, 68–73. [Google Scholar] [CrossRef]
  31. Cortellessa, G.; De Benedictis, R.; Fracasso, F.; Orlandini, A.; Umbrico, A.; Cesta, A. AI and robotics to help older adults: Revisiting projects in search of lessons learned. Paladyn J. Behav. Robot. 2021, 12, 356–378. [Google Scholar] [CrossRef]
  32. Wang, S.; Bolling, K.; Mao, W.; Reichstadt, J.; Jeste, D.; Kim, H.-C.; Nebeker, C. Technology to support aging in place: Older adults’ perspectives. Healthcare 2019, 7, 60. [Google Scholar] [CrossRef] [PubMed]
  33. Chan, M.; Estève, D.; Fourniols, J.-Y.; Escriba, C.; Campo, E. Smart wearable systems: Current status and future challenges. Artif. Intell. Med. 2012, 56, 137–156. [Google Scholar] [CrossRef] [PubMed]
  34. Chen, K.; Chan, A.H.-S. Use or non-use of gerontechnology—A qualitative study. Int. J. Environ. Res. Public Health 2013, 10, 4645–4666. [Google Scholar] [CrossRef]
  35. Chen, K.; Chan, A.H.S. Gerontechnology acceptance by elderly Hong Kong Chinese: A senior technology acceptance model (STAM). Ergonomics 2014, 57, 635–652. [Google Scholar] [CrossRef]
  36. Zwijsen, S.A.; Niemeijer, A.R.; Hertogh, C.M. Ethics of using assistive technology in the care for community-dwelling elderly people: An overview of the literature. Aging Ment. Health 2011, 15, 419–427. [Google Scholar] [CrossRef]
  37. Park, J.; Han, S. Investigating older consumers’ acceptance factors of autonomous vehicles. J. Retail. Consum. Serv. 2023, 72, 103241. [Google Scholar] [CrossRef]
  38. Li, C.; Lee, C.-F.; Xu, S. Stigma Threat in Design for Older Adults: Exploring Design Factors that Induce Stigma Perception. Int. J. Des. 2020, 14, 51–64. [Google Scholar]
  39. Di Giacomo, D.; Ranieri, J.; D’Amico, M.; Guerra, F.; Passafiume, D. Psychological barriers to digital living in older adults: Computer anxiety as predictive mechanism for technophobia. Behav. Sci. 2019, 9, 96. [Google Scholar] [CrossRef]
  40. Homburg, C.; Schwemmle, M.; Kuehnl, C. New product design: Concept, measurement, and consequences. J. Mark. Anal. 2015, 79, 41–56. [Google Scholar] [CrossRef]
  41. Huang, K.-L.; Duan, S.-F.; Lyu, X. Affective Voice Interaction and Artificial Intelligence: A research study on the acoustic features of gender and the emotional states of the PAD model. Front. Psychol. 2021, 12, 664925. [Google Scholar] [CrossRef] [PubMed]
  42. Yang, B.; Liu, Y.; Liang, Y.; Tang, M. Exploiting user experience from online customer reviews for product design. Int. J. Inf. Manag. 2019, 46, 173–186. [Google Scholar] [CrossRef]
  43. Menghi, R.; Papetti, A.; Germani, M. Product Service Platform to improve care systems for elderly living at home. Health Policy Technol. Soc. 2019, 8, 393–401. [Google Scholar] [CrossRef]
  44. Demirbilek, O.; Demirkan, H. Universal product design involving elderly users: A participatory design model. Appl. Ergon. 2004, 35, 361–370. [Google Scholar] [CrossRef]
  45. Jarrahi, M.H. Artificial intelligence and the future of work: Human-AI symbiosis in organizational decision making. Bus. Horiz. 2018, 61, 577–586. [Google Scholar] [CrossRef]
  46. Jones, C.T.; Rice, C.; Lam, M.; Chandler, E.; Lee, K.J. Toward TechnoAccess: A narrative review of disabled and aging experiences of using technology to access the arts. Technol. Soc. 2021, 65, 101537. [Google Scholar] [CrossRef]
  47. Belk, R. Ethical issues in service robotics and artificial intelligence. Serv. Ind. J. 2021, 41, 860–876. [Google Scholar] [CrossRef]
  48. Scherer, M.U. Regulating artificial intelligence systems: Risks, challenges, competencies, and strategies. Harv. JL Tech. 2015, 29, 353. [Google Scholar] [CrossRef]
  49. Long, D.; Magerko, B. What is AI literacy? Competencies and design considerations. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, Honolulu, HI, USA, 25–30 April 2020; pp. 1–16. [Google Scholar]
  50. Cukier, K. Commentary: How AI shapes consumer experiences and expectations. J. Mark. Res. 2021, 85, 152–155. [Google Scholar] [CrossRef]
  51. Meiselman, H.L. A review of the current state of emotion research in product development. Food Res. Int. 2015, 76, 192–199. [Google Scholar] [CrossRef]
  52. Khalid, H.M.; Helander, M.G. Customer emotional needs in product design. Concurr. Eng. 2006, 14, 197–206. [Google Scholar] [CrossRef]
  53. Corrigan, P.W.; Morris, S.B.; Michaels, P.J.; Rafacz, J.D.; Rüsch, N. Challenging the public stigma of mental illness: A meta-analysis of outcome studies. Psychiatr. Serv. 2012, 63, 963–973. [Google Scholar] [CrossRef] [PubMed]
  54. Dogra, N.; Adil, M.; Sadiq, M.; Rafiq, F.; Paul, J. Demystifying tourists’ intention to purchase travel online: The moderating role of technical anxiety and attitude. Curr. Issues Tour. 2022, 26, 2164–2183. [Google Scholar] [CrossRef]
  55. Gan, Y.; Ji, Y.; Jiang, S.; Liu, X.; Feng, Z.; Li, Y.; Liu, Y. Integrating aesthetic and emotional preferences in social robot design: An affective design approach with Kansei engineering and deep convolutional generative adversarial network. Int. J. Ind. Ergon. 2021, 83, 103128. [Google Scholar] [CrossRef]
  56. Chung, J.; Demiris, G.; Thompson, H.J. Ethical considerations regarding the use of smart home technologies for older adults: An integrative review. Annu. Rev. Nurs. Res. 2016, 34, 155–181. [Google Scholar] [CrossRef]
  57. Ploug, T.; Holm, S. The right to refuse diagnostics and treatment planning by artificial intelligence. Med. Health Care Philos. 2020, 23, 107–114. [Google Scholar] [CrossRef]
  58. Lachman, M.E.; Weaver, S.L. The sense of control as a moderator of social class differences in health and well-being. J. Personal. Soc. Psychol. 1998, 74, 763–773. [Google Scholar] [CrossRef]
  59. Xu, H.; Gupta, S.; Rosson, M.B.; Carroll, J.M. Measuring mobile users’ concerns for information privacy. In Proceedings of the Thirty Third International Conference on Information Systems, Orlando, FL, USA, 16–19 December 2012; pp. 1–16. [Google Scholar]
  60. Vogel, D.L.; Bitman, R.L.; Hammer, J.H.; Wade, N.G. Is stigma internalized? The longitudinal impact of public stigma on self-stigma. J. Couns. Psychol. 2013, 60, 311. [Google Scholar] [CrossRef]
  61. Jeng, M.-Y.; Pai, F.-Y.; Yeh, T.-M. Antecedents for Older Adults’ Intention to Use Smart Health Wearable Devices-Technology Anxiety as a Moderator. Behav. Sci. 2022, 12, 114. [Google Scholar] [CrossRef]
  62. Lin, C.S.; Wu, S.; Tsai, R.J. Integrating perceived playfulness into expectation-confirmation model for web portal context. Inf. Manag. 2005, 42, 683–693. [Google Scholar] [CrossRef]
  63. Ogbanufe, O.; Gerhart, N. Exploring smart wearables through the lens of reactance theory: Linking values, social influence, and status quo. Comput. Hum. Behav. 2022, 127, 107044. [Google Scholar] [CrossRef]
  64. Belanche, D.; Casaló, L.V.; Flavián, C. Artificial Intelligence in FinTech: Understanding robo-advisors adoption among customers. Ind. Manag. Data Syst. 2019, 119, 1411–1430. [Google Scholar] [CrossRef]
  65. Cristea, M.; Noja, G.G.; Stefea, P.; Sala, A.L. The impact of population aging and public health support on EU labor markets. Int. J. Environ. Res. Public Health 2020, 17, 1439. [Google Scholar] [CrossRef] [PubMed]
  66. Tian, H.; Iqbal, S.; Akhtar, S. Exploring predictors of innovation performance of SMEs: A PLS-SEM approach. Empl. Relat. Int. J. 2023, 45, 909–924. [Google Scholar] [CrossRef]
  67. Nguyen, P.-H.; Tsai, J.-F.; Lin, M.-H.; Hu, Y.-C. A hybrid model with spherical fuzzy-ahp, pls-sem and ann to predict vaccination intention against COVID-19. Mathematics 2021, 9, 3075. [Google Scholar] [CrossRef]
  68. Shrestha, N. Detecting multicollinearity in regression analysis. Am. J. Appl. Math. Stat. 2020, 8, 39–42. [Google Scholar] [CrossRef]
  69. Podsakoff, P.M.; Organ, D.W. Self-reports in organizational research: Problems and prospects. J. Manag. 1986, 12, 531–544. [Google Scholar] [CrossRef]
  70. Mattila, A.S.; Enz, C.A. The role of emotions in service encounters. J. Serv. Res. 2002, 4, 268–277. [Google Scholar] [CrossRef]
  71. Chin, W.W.; Thatcher, J.B.; Wright, R.T. Assessing common method bias: Problems with the ULMC technique. MIS Q. 2012, 36, 1003–1019. [Google Scholar] [CrossRef]
  72. Hair, J.F.; Sarstedt, M.; Ringle, C.M.; Mena, J.A. An assessment of the use of partial least squares structural equation modeling in marketing research. J. Acad. Mark. Sci. 2012, 40, 414–433. [Google Scholar] [CrossRef]
  73. Hair, J.F., Jr.; Sarstedt, M.; Ringle, C.M.; Gudergan, S.P. Advanced Issues in Partial Least Squares Structural Equation Modeling; SAGE Publications: Los Angeles, CA, USA, 2017. [Google Scholar]
  74. Etemad-Sajadi, R.; Gomes Dos Santos, G. Senior citizens’ acceptance of connected health technologies in their homes. Int. J. Health Care Qual. Assur. 2019, 32, 1162–1174. [Google Scholar] [CrossRef] [PubMed]
  75. Essén, A. The two facets of electronic care surveillance: An exploration of the views of older people who live with monitoring devices. Soc. Sci. Med. 2008, 67, 128–136. [Google Scholar] [CrossRef] [PubMed]
  76. Choudrie, J.; Junior, C.-O.; McKenna, B.; Richter, S. Understanding and conceptualising the adoption, use and diffusion of mobile banking in older adults: A research agenda and conceptual framework. J. Bus. Res. 2018, 88, 449–465. [Google Scholar] [CrossRef]
  77. Kotler, P.; Kartajaya, H.; Setiawan, I. Marketing 5.0: Technology for Humanity; John Wiley & Sons: Hoboken, NJ, USA, 2021. [Google Scholar]
  78. Nimrod, G. Technostress: Measuring a new threat to well-being in later life. Aging Ment. Health 2018, 22, 1086–1093. [Google Scholar] [CrossRef]
  79. Guido, G.; Ugolini, M.M.; Sestino, A. Active ageing of elderly consumers: Insights and opportunities for future business strategies. SN Bus. Econ. 2022, 2, 8. [Google Scholar] [CrossRef]
  80. Moschis, G.P. Consumer behavior in later life: Current knowledge, issues, and new directions for research. Psychol. Mark. 2012, 29, 57–75. [Google Scholar] [CrossRef]
Systems 11 00551 g001
Figure 1. Research model.
Figure 1. Research model.
Systems 11 00551 g001
Systems 11 00551 g002
Figure 2. Model results. *** p < 0.001; ** p < 0.01.
Figure 2. Model results. *** p < 0.001; ** p < 0.01.
Systems 11 00551 g002
Table 1. Negative effects of AIPs on older people.
Table 1. Negative effects of AIPs on older people.
Research ObjectsResearch SubjectsNegative EffectsReference
1ICT and older peopleThe impact of digital participation on the quality of life of older peoplePerceived control; feelings of shame; privacy disclosure; social isolation[28]
2AIThe ethical issues of geriatric technology in elderly careDiscrimination; dehumanization[5]
3AI and IoTThe influence of AI on life assistance and health monitoring of older peoplePerceived control; perceived intrusiveness[23]
4AIOpportunities and challengesPrejudice; discrimination[6]
5AI and expert systemsExpectations of AIExpectation disconfirmation[29]
6Virtual personal assistantActive agingPerceived intrusiveness[30]
7AI and roboticsLessons from intelligent products for older peopleNon-availability; emotional reaction; discrimination; loss of autonomy[31]
8AIAging in placePerceived control; non-availability; lack of technical literacy[32]
9Intelligent wearable systemStatus and challengesStigma; feelings of shame; perceived intrusiveness[33]
10Intelligent assistive technologyThe emotional experiences and attitudesPerceived control; lack of literacy; stigma; feelings of shame[15]
11Geriatric technologyReasons for negative behavior of older peopleSocial isolation; addiction[34]
12Geriatric technologyThe acceptance of geriatric technologyTechnology anxiety[35]
13Assistive
equipment		The ethical discussion of technologies in the communityStigma; feelings of shame; private anxiety; perceived control; perceived intrusiveness[36]
14Autonomous vehiclesExternal and internal factors for acceptanceStigma; stereotype[37]
15Wearable devices and sensorsQuantified self Stigma; feelings of shame; perceived control[38]
16Advanced technology (AI and robotics)Psychological barriers to digital societyTechnology anxiety[39]
17Assistive technologyBarriers to technology adoption Perceived uselessness; stigma;
not being independent		[4]
Notes: ICT = information and communications technology; IoT = Internet of Things.
Table 2. Variables and indicators.
Table 2. Variables and indicators.
VariableMeasurement ItemsReference
Perceived ControlPC1: I feel like I’m losing the territory that I used to control.[58]
PC2: I feel like I lack control over the outside world (other people, situations).
PC3: I can set clear, realistic, and meaningful goals.
PC4: Something (human or machine) exerts too much control over me.
Perceived IntrusivenessPI1: I am concerned that AIPs are collecting too much information about me.[59]
PI2: I feel that as a result of my using an AIP, others know about me more than I am comfortable with.
PI3: I believe that as a result of my using an AIP, information about me that I consider private is now more readily available to others than I would want.
PI4: I feel that as a result of my using an AIP, information about me is out there that, if used, will invade my privacy.
Self-StigmaSS1: It makes me feel inferior to use an AIP.[60]
SS2: When I use an AIP, my view of myself is more negative.
SS3: My self-image feels threatened when I use an AIP.
SS4: Using an AIP makes me feel like there is something wrong with me.
Public StigmaPS1: Using an AIP carries a social stigma.
PS2: It is a sign of weakness and aging to use an AIP.
PS3: People tend to like others less when those others are using an AIP.
PS4: It is advisable for me to hide that I use an AIP.
Socio-emotional MismatchSM1: AIPs cannot satisfy my emotional needs.[15]
SM2: AIPs cannot match my emotional needs.
SM3: I cannot say that AIPs please me.
SM4: AIPs have no positive impact on my affection.
Functional MismatchFM1: AIPs can not meet my daily needs.
FM2: AIPs don’t fit my daily needs.
FM3: I cannot say that AIPs help me in my life.
FM4: AIPs have not changed my life.
Expectation DisconfirmationED1: My experience with using the AIP was worse than what I expected.[62]
ED2: The service level provided by the AIP was worse than what I expected.
ED3: Overall, most of my expectations about using the AIP were not confirmed.
Technology AnxietyTA1: I feel stressed when I use a new AIP.[61]
TA2: I am worried that the new AIP will affect my life.
TA3: I fear that AIPs will change my life.
TA4: I’m afraid that I don’t have enough ability to use AIPs.
Avoidance BehaviorAB1: The transition to AIPs is stressful for me.[63]
AB2: I feel comfortable not continuing to use AIPs.
AB3: I like using the original product instead of AIPs.
Exit BehaviorEB1: I won’t be using the AIPs as much as I used to.
EB2: After using an AIP for a while, my interest in continuing to use it gradually decreases.
EB3: I’m going to stop using my AIPs, but that doesn’t mean I’m going to give them up altogether.
Table 3. Respondents’ sample statistics.
Table 3. Respondents’ sample statistics.
MeasureItemCount
Age55–59212 (19.24%)
60–69515 (46.73%)
70–79237 (21.50%)
>80138 (12.53%)
GenderMale585 (53.09%)
Female517 (46.91%)
EducationPrimary school116 (10.53%)
Junior middle school477 (43.28%)
High school353 (32.03%)
Undergraduate156 (14.16%)
AI used (multi-choice)Healthy670 (60.80%)
Accompanied784 (71.14%)
Monitored836 (78.86%)
Walking-aided539 (48.91%)
Table 4. Test of fitting validity.
Table 4. Test of fitting validity.
IndexValueHI95Result
SRMR0.0340.133Support
d_ULS0.82812.439Support
d_G0.590.938Support
Table 5. Reliability and validity.
Table 5. Reliability and validity.
ConstructItemLoadingCronbach’s αCRAVE
Function MismatchFM10.9410.9440.9600.857
FM20.941
FM30.943
FM40.877
Avoidance BehaviorAB10.8970.9080.9420.845
AB20.939
AB30.922
Socio-emotion MismatchSM10.9280.9620.9720.898
SM20.960
SM30.954
SM40.947
Technology AnxietyTA10.9210.9340.9530.835
TA20.905
TA30.912
TA40.917
Expectation DisconfirmationED10.8570.8610.9160.783
ED20.912
ED30.885
Public StigmaPS10.9490.9600.9710.892
PS20.949
PS30.960
PS40.920
Perceived IntrusivenessPI10.7790.8840.9200.743
PI20.869
PI30.898
PI40.897
Perceived ControlPC10.9180.9500.9640.869
PC20.928
PC30.948
PC40.935
Self-stigmaSS10.9250.9470.9620.863
SS20.936
SS30.920
SS40.935
Exit BehaviorEB10.7970.8270.8970.744
EB20.886
EB30.902
Table 6. Fornell–Larcker criterion.
Table 6. Fornell–Larcker criterion.
FMABSMTAEDPSPIPCSSEB
FM0.926
AB0.6250.919
SM0.8510.6170.947
TA0.6100.5560.6270.914
ED0.5810.5380.5210.4590.885
PS0.7770.6160.7500.6970.5410.945
PI0.5110.5060.4710.5020.7690.5100.862
PC0.6000.5670.5550.5160.7540.5740.7070.932
SS0.6020.5950.5540.5340.7230.5720.6770.8420.929
EB0.6880.5580.6050.4520.5190.6450.4570.5400.5370.863
Notes: FM = Functional Mismatch; PI = Perceived Intrusiveness; PC = Perceived Control; ED = Expectation Disconfirmation; SM = Socio-emotional Mismatch; SS = Self-Stigma; PS = Public Stigma; TA = Technology Anxiety; AB = Avoidance Behavior; EB = Exit Behavior.
Table 7. HTMT.
Table 7. HTMT.
FMABSMTAEDPSPIPCSSEB
FM
AB0.675
SM0.8420.660
TA0.6440.5990.655
ED0.6440.6080.5720.510
PS0.8160.6600.7800.7280.595
PI0.5550.5630.5070.5490.8290.551
PC0.6340.6100.5800.5450.8340.6010.767
SS0.6360.6410.5800.5650.8010.5990.7340.837
EB0.7790.6470.6780.5120.6160.7240.5340.6100.608
Notes: FM = Functional Mismatch; PI = Perceived Intrusiveness; PC = Perceived Control; ED = Expectation Disconfirmation; SM = Socio-emotional Mismatch; SS = Self-Stigma; PS = Public Stigma; TA = Technology Anxiety; AB = Avoidance Behavior; EB = Exit Behavior.
Table 8. Results of hypotheses testing.
Table 8. Results of hypotheses testing.
HypothesesPath CoefficientT Valuep-ValueResults
H1: PI -> FM0.0431.020.307No support
H2: PC -> FM0.3637.842<0.001Support
H3: ED -> FM0.2756.071<0.001Support
H4: SS -> SM0.1544.441<0.001Support
H5: PS ->SM0.54912.610<0.001Support
H6: TA -> SM0.1623.5440.001Support
H7: FM -> AB0.3635.683<0.001Support
H8: FM -> EB0.63012.237<0.001Support
H9: SM -> AB0.3084.902<0.001Support
H10: SM -> EB0.0691.3070.191No support
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Hong, W.; Liang, C.; Ma, Y.; Zhu, J. Why Do Older Adults Feel Negatively about Artificial Intelligence Products? An Empirical Study Based on the Perspectives of Mismatches. Systems 2023, 11, 551. https://doi.org/10.3390/systems11110551

AMA Style

Hong W, Liang C, Ma Y, Zhu J. Why Do Older Adults Feel Negatively about Artificial Intelligence Products? An Empirical Study Based on the Perspectives of Mismatches. Systems. 2023; 11(11):551. https://doi.org/10.3390/systems11110551

Chicago/Turabian Style

Hong, Wenjia, Changyong Liang, Yiming Ma, and Junhong Zhu. 2023. "Why Do Older Adults Feel Negatively about Artificial Intelligence Products? An Empirical Study Based on the Perspectives of Mismatches" Systems 11, no. 11: 551. https://doi.org/10.3390/systems11110551

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop