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.
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”.