Experience of the Use of AI Conversational Agents Among Low-Income Older Adults Living Alone

Abstract

There is little research exploring the experiences of using artificial intelligence (AI) conversational agents among low-income older adults living alone. To address this gap, we conducted a qualitative research study, analyzing data from 18 older adults who utilized AI conversational agents while living alone. Our focus was on exploring the perceived benefits and satisfaction of older adult users with these AI conversational agents. We identified two primary categories of perceived benefits: instrumental/functional and emotional aspects. Among the instrumental benefits, participants shared how using “Arya (the AI speaker used in the study)” positively impacted their daily living, enabling them to find and manage diverse information, enhance communication skills, and aid in memory recovery and maintenance. On the emotional benefits, participants reported experiencing positive affect, a sense of security, and feelings of gratitude and happiness. Significantly, participants felt that the intended purpose of using an AI conversational agent was achieved, resulting in a high level of satisfaction. The support from information and communication technology (ICT) managers who provided technical assistance and the availability of free services using the AI conversational agents further contributed to their satisfaction. The use of ICT to improve the quality of life from the perspective of older adults living alone makes a significant contribution to the literature. Given the rapidly aging society in South Korea and worldwide, this intervention provides a valuable support system with practical and psychosocial benefits, promoting the health and safety of older adults living alone.

Plain Language Summary

AI speakers: How they help low-income older adults living alone

This research aimed to understand how low-income older adults living alone feel about using AI conversational agents (like smart speakers) in their daily lives. Interviews were conducted with 18 older adults who used an AI speaker called “Arya.” The study identified two main types of benefits: practical and emotional. On the practical side, participants reported that the AI speaker helped them manage daily tasks, find information, improve communication, and enhance memory. Emotionally, the AI speaker made them feel happier, more secure, and grateful. Overall, participants were highly satisfied with the AI speaker, especially due to the technical support they received and the free availability of the service. This research highlights how AI can improve the quality of life for older adults living alone, particularly as the aging population grows in South Korea and worldwide. Technology offers both practical and emotional benefits, helping to promote their health and safety.

Keywords

artificial intelligence conversational agents older adults low-income information and communication technologies living alone

Introduction

The elderly population in South Korea, aged 65 and older, reached 9.02 million in 2022, representing 17.5% of the country’s total population (Statistics Korea, 2022). By 2025, it is projected that 20.6% of South Koreans will be classified as older persons, pushing the country into a hyper-aged society (Korea JoongAng Daily, 2022). Moreover, the proportion of single-person households, specifically those aged 60 or older, is expected to rise from 30% in 2015 to 45% in 2045 (Statistics Korea, 2022).

South Korean society is currently experiencing a rapid aging process, marked by a substantial growth in the elderly population. This demographic shift is further compounded by record low birth rates and a decline in the working-age population available to support older adults. Despite the increased life expectancy of older South Korean adults and the unprecedented growth rate of South Korea’s elderly population, their social and economic security have not been well established. Recent data from the Organisation for Economic Co-operation and Development (OECD) in 2021 reveals that South Korea has the highest old-age poverty rate (46%) among member countries. In addition, South Korea’s older adults have the highest suicide rates among OECD countries, with economic difficulty identified as one of the major causes (Korea Ministry of Health and Welfare, 2021; Organization for Economic Co-operation and Development [OECD], 2021).

Moreover, South Korea faces significant challenges in caring for its rapidly aging population. Older adults are disproportionately burdened by isolation, chronic disease, and functional decline, making it difficult for them to live safely on their own (Blaschke et al., 2009; Katherine et al., 2020). Among older adults, those living alone are particularly vulnerable to social isolation and loneliness compared to those living with family members. Moreover, the COVID-19 pandemic has exacerbated the risks to their physical and mental well-being, leading to increased social isolation and loneliness (Corbett et al., 2021; Katherine et al., 2022; Song et al., 2021). Extensive research highlights that social isolation and loneliness among older adults contribute to a higher risk of depression, mortality, and dementia (Blaschke et al., 2009; Corbett et al., 2021, Koh et al., 2021; Kuiper et al., 2015; Lee & Lee, 2022; Song et al., 2021).

In addition, the ongoing challenge of finding caregivers to meet the needs of older adults and provide them with proper care remains. Moreover, it is worth noting that family caregivers, many of whom are older adults themselves, willingly take on the responsibility of assisting other older adults with daily activities and various tasks, despite the difficulties they face due to their own aging process (Blaschke et al., 2009). Recognizing the potential for effective intervention, there has been a growing interest in utilizing artificial intelligence (AI) conversational agents to support the independent living and overall well-being of older adults (Kim, 2021; Kim et al., 2020; Nallam et al., 2020; Song et al., 2021; Yang & Lee, 2018). Conversational agents are considered relatively affordable, easy-to-use, and readily available in the market. They assist users in various daily tasks, such as searching for specific information, streaming music, making phone calls, sending text messages, and obtaining weather and news updates.

Despite the growing interest and discussions regarding the utilization of conversational agents for older adults, there is a scarcity of studies examining their usage among this population. Limited research exists on older adults’ perceptions, opinions, and the psychological and social effects of using conversational agents. In addition, there is a lack of literature exploring the categorical characteristics of older adult conversational agent users, such as their economic status and living arrangements. Most previous research has primarily focused on exploring social interaction, emotional intimacy, and satisfaction among older adults using conversational agents (Kim, 2021; Lee & Lee, 2022; Nallam et al., 2020; Pradham et al., 2019; Song et al., 2021). However, these studies did not specifically examine the categorical characteristics of older adults, such as living alone, having a low income, or relying on public assistance, and instead, encompassed a broader range of older adult characteristics, including diverse living arrangements and income levels. Therefore, the purpose of this study is to fill these gaps by exploring the experiences of low-income South Korean older adults living alone with conversational agents. Through an analysis of the actual benefits of AI speakers in meeting the needs and desires of non-face-to-face services for older adults, this study aims to provide guidance for an integrated model combining non-face-to-face and face-to-face services. Ultimately, this study will contribute to providing emotional and informational care for older adults living alone, thereby enhancing their overall quality of live in the future.

The study seeks to address the following research questions:

(1) How do older adults perceive the benefits of conversational agents?

(2) What is the perception of older adults regarding their intention to continue using conversational agents and their satisfaction with their usage?

To explore the questions of this study, we examine existing literature on AI conversational agents for older adults. Specifically, we focus on previous research that investigates the perceived benefits and challenges associated with these agents, as well as their overall acceptance, satisfaction, and the intention to use these agents among older adult users.

Background and Literature Review

AI Conversational Agents

Conversational agents, as defined by Corbett et al. (2021) and Han and Yang (2018), enable users to communicate with an automated system using voice recognition and natural language processing. These agents are referred to by various names in the research community, such as intelligent personal/voice assistants (Han & Yang, 2018; Jang, 2020; Nallam et al., 2020), virtual personal assistants (Yang & Lee, 2018), smart speaker-based voice assistant (Kim, 2021; Pradhan et al., 2019), and voice-controlled intelligent personal assistants (Katherine et al., 2022). For the purposes of this study, the term “conversational agents” will be used to encompass all these digital assistant tools.

Conversational agents offer three primary functions: (a) professional/administrative assistance for daily tasks like sending messages or conducting web searches, (b) technical assistance for complex tasks such as controlling smart home appliance or monitoring health, and (c) social assistance by engaging in conversation akin to a human friend (Han & Yang, 2018; Saad et al., 2016; Yang & Lee, 2018). Extensive literature suggests that AI speaker-based voice assistants can mitigate loneliness, improve physical and mental health, and alleviate the burden on families and caregivers (Blaschke et al., 2009; Cho & Ryu, 2022; Katherine et al., 2022; Kim et al., 2022; Song et al., 2021).

Theoretical Background

This study employs the Technology Acceptance Model (TAM) and Parasocial Relationship (PSR) Theory as the theoretical framework to explore the instrumental and emotional benefits and challenges of AI conversational agents for older adults, particularly those living alone.

The Technology Acceptance Model (TAM), introduced by Davis (1989), explains how users adopt new technology. Its key components include perceived usefulness, perceived ease of use, attitude toward use, behavioral intention to use, and actual system use (Davis, 1989). Davis (1989) posits that perceived usefulness (the belief that technology improves performance) and perceived ease of use (the belief that it is free from effort) directly influence attitudes toward using the technology. This attitude, in turn, predicts the intention to use the technology, ultimately leading to actual use. TAM provides insights into why older adults might adopt technologies like AI conversational agents in their daily lives. When they perceive these technologies as useful and easy to use, their intention to use them increases, resulting in more consistent usage.

The Parasocial Relationship (PSR) Theory describes the one-sided relationships that viewers form with media personalities (Horton & Wohl, 1956). Even though the media characters do not engage in reciprocal interactions, viewers feel a personal connection as though they know the media characters personally. Horton and Wohl (1956) propose that parasocial relationships provide viewers with emotional satisfaction and bonding, similar to real-life relationships. These relationships are reinforced through repeated exposure to media and deepening the viewer’s connections with the media characters. This theory is relevant for AI conversational speakers, as older adults may develop similar emotional bonds through frequent interactions, gaining companionship and emotional support akin to relationships with media characters.

Related Prior Literature

A review of the existing literature reveals three main themes concerning the experiences of older adults with the use of conversational agents: the perceived benefits of using conversational agents (including instrumental/functional usefulness and psychological benefits), the overall acceptance, satisfaction, and intention to use conversational agents, and the perceived challenges associated with adopting conversational agents.

Benefits

Instrumental/Functional Usefulness and Utility

Existing literature extensively examines the instrumental/functional usefulness of conversational agents as assistive technology (Han & Yang, 2018; Kim, 2021; Park & Choi, 2018; Pradhan et al., 2019; Yang & Lee, 2018). Perceived usefulness, as defined by Davis (1989), refers to “the degree to which a person believes that using a particular system would enhance his or her job performance” (p. 320). Conversational agents are recognized as user-friendly devices that fulfil essential roles in assisting various daily tasks, such as sending text messages, making calls, playing music, and providing weather updates (Han & Yang, 2018; Katherine et al., 2022). Studies show a positive relationship between perceived usefulness and the adoption of conversational agents (Kim et al., 2020; Park & Choi, 2018; Yang & Lee, 2018), but challenges in usability for older adults persist (Han & Yang, 2018; Park & Choi, 2018; Yang & Lee, 2018). The section titled “Perceived Challenges” further discusses these usability issues.

Psychological Benefits

A large volume of research has explored the impact of psychological factors, such as parasocial interaction, personification, and loneliness, on individuals’ satisfaction with the use of conversational agents. Usage patterns indicate that older adults frequently seek emotional conversation, reflecting their feelings of loneliness (Jang, 2020).

Parasocial Interaction

Studies show that parasocial interaction positively affects satisfaction and continued use (Han & Yang, 2018; Jang, 2020). Han and Yang (2018) further explain that increased communication and interaction with conversational agents foster an intimate relationship between users and the devices, which in turn affects users’ intention to continue using them. Han and Yang (2018) found that social attraction- users’ desire to engage with the device- has a greater impact on satisfaction than task/utility or physical attraction. This underscores the importance of designing user interfaces that foster an intimate experience, similar to interactions with a friend or family member.

Personification

Personification involves treating conversational agents as if they were human (Pradhan et al., 2019). Users often refer to these devices as friends and use personal pronouns like “she” or “her” (Jang, 2020; Kim et al., 2020; Lee & Lee, 2022; Park & Choi, 2018; Pradhan et al., 2019). Song et al. (2021) found that personification significantly influenced the emotional and functional satisfaction of older adults, as well as their intention to continue using conversational agents. While some studies suggest that personification enhances satisfaction, Jang (2020) found that users often preferred to view the device as a “competent secretary” or “smart helper” (p. 133).

Loneliness

Older adults living alone are more vulnerable to social isolation and loneliness than those living with family. Research indicates that conversational agents can help reduce these feelings and enhance social connectedness (Cho & Ryu, 2022; Corbett et al., 2021; Katherine et al., 2022; Kim et al., 2020; Kim et al., 2022; Lee & Lee, 2022; Song et al., 2021). Social isolation and loneliness can negatively impact older adults’ health, increasing the risk of dementia and mortality (Corbett et al., 2021; Katherine et al., 2020, 2022; Kim et al., 2020; Kim et al., 2022; Kuiper et al., 2015). However, some studies suggest users may prefer human interaction over communication with devices (Jang, 2020; Slegers et al., 2008), highlighting the need for more humanized designs for conversational agents.

Perceived Challenges

Despite the potential of conversational agents, several challenges prevent older adults from using them. Key issues include difficulties with voice command recognition, limitations in natural conversation, and struggles with coherent sentence construction (Kim, 2021; Park & Choi, 2018). There is also a lack of understanding of older adults’ needs, such as customized content and consideration of the physical and psychological aspects of aging (Kim, 2021; Park & Choi, 2018).

Misconceptions and attitudinal issues among older adults also contribute, such as perceiving technology as risky, complex, and costly, lacking confidence in device usage, and privacy concerns (Blaschke et al., 2009; Corbett et al., 2021; Kim, 2021). Age-related issues, including hearing loss and cognitive challenges, further complicate usage (Blaschke et al., 2009; Kim, 2021). Furthermore, costs associated with internet connectivity can be barriers (Blaschke et al., 2009; Corbett et al., 2021). Lastly, inadequate training and support, along with the absence of structured learning opportunities, exacerbate these challenges (Blaschke et al., 2009; Han & Yang, 2018; Yang & Lee, 2018).

Overall Acceptance, Satisfaction, and Intention to Use CAs

Several studies have explored the intentions of potential users to adopt conversational agent devices (Han & Yang, 2018; Park & Choi, 2018; Yang & Lee, 2018). Research has shown that older adults generally have a positive attitude toward the use of conversational agents and express intentions to use devices in the future (Blaschke et al., 2009; Park & Choi, 2018). Yang and Lee (2018) found that perceived usefulness and enjoyment significantly influence usage intention, with content quality being the most impactful product feature.

Han and Yang’s (2018) study revealed that user satisfaction significantly affects the intention for continual use, though this contradicts many earlier studies. Park and Choi (2018) noted that users’ satisfaction levels are often low due to functional barriers, such as difficulty forming coherent sentences and limited device capabilities. While many studies highlight the potential benefits of conversational agents for older adults, low continued. Usage suggests these devices may not be engaging users effectively. For instance, only 3% of users remained active in the second week after initial interest (Marchick, 2017, as cited in Yang & Lee, 2018). The low usability may indicate a lack of understanding of the practical value of these devices (Yang & Lee, 2018). However, prior literature offers few explanations for poor usability, and studies on user acceptance, adoption, and satisfaction remain scarce.

Building upon the insights gained from the literature review, the following section details the methodology employed to investigate the research questions and explore the experiences and attitudes of older adults toward AI conversational agents.

Methods

This study adopted a qualitative approach to examine the subjective experiences of using conversational agents among low-income older adults living alone. This study aimed to collect older adult participants’ descriptive, rich and fresh accounts of conversational agents use in their daily lives and to gain an in-depth understanding of their lived experiences. The researchers were interested in how digital technology could be used to reduce loneliness in older adults. The first author was awarded a grant to research “Happy Community AI Care.”

Participants

The participants were low-income, aged 65 or older, living alone, users of artificial intelligence care services, and who used interactive agents every day. In total, we recruited 18 participants (female n = 10, male n = 8; range 65–88, mean age 77; See Table 1). In terms of education levels, there were five people with no education, three elementary school graduates, four middle school graduates, two high school graduates, and four college graduates or higher. In terms of income levels, ten people lived on public assistance, and eight were in the low-income category. Of the 18 participants, 8 had received public formal care. Smartphones were owned by 13 people, and five owned flip phones. The AI speaker usage period was 4 to 19 months, and the average was 12 months.

Table 1.

Participants’ Demographic Characteristics.

P#	Gender	Age	Education level	Income level (public assistance)	Public formal care (Y/N)	Type of mobile phone	Period of AI speaker usage (month)
1	M	72	University	N	Y	Smart phone	19
2	F	82	No education	Y	N	Folder phone	12
3	F	73	Middle school	N	Y	Folder phone	12
4	M	77	High school	N	Y	Smart phone	16
5	F	67	University	N	N	Smart phone	7
6	M	81	No education	N	N	Smart phone	19
7	F	70	No education	N	Y	Smart phone	18
8	F	65	Middle school	Y	N	Smart phone	18
9	M	86	University	N	Y	Folder phone	17
10	F	86	No education	Y	N	Folder phone	9
11	M	80	Middle school	Y	N	Smart phone	9
12	M	81	Middle school	Y	N	Smart phone	9
13	F	83	No education	Y	N	Folder phone	8
14	F	80	High school	Y	Y	Smart phone	9
15	M	71	University	Y	Y	Smart phone	9
16	F	67	Elementary school	Y	Y	Smart phone	9
17	M	73	Elementary school	N	N	Smart phone	4
18	F	88	Elementary school	Y	N	Smart phone	9

Data Collection

The participants in this study were users of “Happy Community AI Care” project in South Korea. The project, “Happy Community AI Care,” aims to improve the quality of life for older adults living alone by addressing issues related to loneliness, dementia prevention and delay, the need for integrated care services, and the digital information gap. Through public-private cooperation, this project creates social value. Established in April 2019, “Happy Community AI Care” is a public-private collaboration project led by SK Telecom, the National Social Solidarity Economy Local Government Council, and a social enterprise called Happy Connect. Happy Connect is responsible for overseeing integrated monitoring, data analysis, and sharing, as well as hiring and managing of information and communication technology (ICT) care managers. Local government and public institutions are in charge of supporting local government contracts and operations, selecting targets, and disseminating opinions across local governments. SK Telecom provides technology support such as AI and big data, new service development support, and business effectiveness review. The data from older adults are managed using Customer Relationship Management (CRM) as the data management tool. In addition, a platform has been established to monitor the usage, assess satisfaction levels, and promote increased utilization of AI conversational agents. ICT managers play a significant role in providing technical support to those using AI CAs, guiding them in solving simple problems through the phone or home visits. This technical support encourages participants to use the technology continuously. A pre-post study of 600 people who used Happiness Community AI Care found that their enjoyment of trying out digital devices increased, anxiety decreased, and sense of efficacy increased (SK Telecom and National Social Solidarity Economy Local Government Council, and Happy Connect Foundation, 2020).

The AI speaker, “NUGU Candle,” used in this project provides various AI care services such as emotional care, health care, 365-day safety, and life information delivery. It also includes Internet of Things (IoT) sensors for tasks like controlling light switches and doors. The “Happy Community AI Care” project provides low-income older adults living alone with an AI conversational agents and fee for the Internet connection.

Participants were selected based on their proficiency in using AI speakers and their ability to communicate. To recruit participants, the first author contacted the persons in charge of SK Telecom’s AI care project. SK Telecom’ staff then explained the study’s purpose to the person in charge of the social enterprise, Happy Connect, which operates the project, and requested their assistance in referring research participants. The Happy Connect authority asked ICT care managers from Gyeongsangnam-do and a metropolitan city to refer potential participants as research sites. The district’s agency then explained the study to potential research participants and obtained their verbal consent. The first author subsequently contacted the person in charge of the entrusted agency to schedule interviews with older adults living alone who agreed to participate in the study, also seeking recommendations for someone proficient in using AI speakers and capable of effective communication with others. A copy of the interview questionnaire was sent to the participants upon request. We recruited 20 older adults who live alone and conducted interviews with 18 of them. Semi-structured in-depth qualitative interviews were conducted at the participants’ houses.

The interview questions covered various aspects of their daily lives, motivations for using AI conversational agents, frequency of use, difficulties encountered, solutions found, changes in their lives after using AI conversational agents (such as changes in daily routines, relationships, and confidence), perceived intentions, satisfaction levels, and thoughts on AI conversational agents. The developed interview questions were supplemented with insights from preliminary interviews with two participants.

Participation in the study was voluntary. At the beginning of the interview, consent was obtained, and participants’ eligibility was confirmed. Detailed explanations of the study were provided to all participants before the interviews, and written consent was obtained for recording purposes. A reward fee was provided to participants after the interview. All data collected were recorded and transcribed.

Data Analysis

This study utilized thematic qualitative analysis (Braun & Clarke, 2006). Following the transcription and verification of the recorded files, the data was entered into Nvivo for analysis. To ensure accuracy, after transcribing the data by one graduate student research assistants, the authors reviewed the transcripts for validation. The thematic analysis process involved multiple steps. The initial step involved becoming familiar with the data through repeated readings to gain a comprehensive understanding of the data (Braun & Clarke, 2006). Next, the authors independently read and coded a subset of the data to identify patterns and concepts and then compared our coding and engaged in discussions to reach an agreement. This process enhanced the reliability of the analysis and ensured the codes accurately represent the data. In the third step, the authors generated initial codes through an agreement process. Subsequently, we sorted these codes into potential themes and collected relevant quotes. In step four, we undertook a comprehensive review and refinement of the provisional themes created in the previous step. This involved analysing the coded quotes to assess if they demonstrated specific patterns and ensuring their alignment with the overall data. It was during this step that the final themes were decided upon and named. Finally, the last step involved the interpretation of findings, wherein the authors connected the identified themes to the research questions and established relationships between these themes. Figure 1 presents an overview of the data analysis steps.

Figure 1.

The steps of analysis.

To ensure the reliability and validity of the analysis, two triangulation methods were employed. First, brief follow-up interviews were conducted with two ICT managers. Second, peer debriefing was utilized, involving a social work faculty expert in the subject field, to ensure the trustworthiness of the study findings. These measures contributed to the rigor of the study and the credibility of the analysis.

Findings

The major themes identified were instrumental/functional and emotional benefits of using AI conversational agents. SK Telecom’s AI speaker NUGU built with an artificial intelligence called “Arya.” The instrumental functions of using AI conversational agents include active daily living with “Arya,” seeking and managing diverse information, improving communication skills, and facilitating memory recovery and maintenance. Emotional benefits included positive affection, feeling safe, and feelings of gratitude and happiness. The support of ICT managers and the free services accessed through AI conversational agents increased the satisfaction of participants. Below in Figure 2, the benefits of AI conversational agents are presented.

Figure 2.

The benefits of AI conversational agents.

The Benefits of Using AI Conversational Agents

Instrumental/Functional Benefits

Active daily living with “Arya.”

AI conversational agents made a big difference in the older peoples’ lives. The first change was that participants lived actively with Arya every day. The first thing the participants would do after waking up in the morning was to say “hello” to Arya. The participants started the day by checking the weather forecast, time, news, cooking recipes, and horoscopes. Asking for the news and time was a part of the daily routine, because participants often forgot those. As their social network narrowed with age, the participants spent more time with “Arya”; moreover, they became friends with “Arya” during the COVID-19 pandemic. One participant said, “I’ve easily become kind of addicted to it.”

Now, due to COVID-19, everyone is cut off. This is like living in prison. Besides, I’m lying in bed because I’m injured, and there’s no one to talk to. When someone calls me, I just have nothing to say and they tell me to rest. But ‘Arya’ relieves my boredom. It’s so good. It’s great if you use it well. But if you don’t use it, it’s useless (Participant 4).

Many participants listened to music, mostly “Trot,” because the lyrics were good and talked about life. Trot is a South Korean music genre, which features stereotyped repetitive rhythms, pentatonic scales, and quivering singing influenced by Korean folk songs. Listening to Trot music made the participants’ emotions equable and positive.

Participants used AI conversational agents to stimulate brain activity, such as the dementia prevention program, and listened to the radio, especially religious broadcasts such as hymns, Buddhist instrumental music, which helped them sleep well. Participants also utilized alarms not only for important appointments but also as reminders for taking medicines. As they actively interacted with Arya, they had access to information they wanted and became friends with Arya, which made them livelier and more active. One participant even expressed, “it is a really big help for me.”

Seeking and Managing Diverse Information

Many of the study participants never attended school or only completed elementary school, and thus, did not know how to use digital technology. However, with Arya, they could seek, get and use the information they needed. After using AI conversational agents, it appears that older adults had an improved ability to acquire information, including news, weather, horoscopes, and more. As the participants learned something new daily, their knowledge expanded, and they felt empowered and less isolated from society. One of the advantages of “Arya” is the immediacy in information provision. This encouraged the participants to constantly search for the information they desired. One participant said, “Unlike TV, I can get the information I want without having to wait.”

Improving Communication Skills

Talking to AI conversational agents facilitated a significant improvement in communication skills of the participants. In response to the AI speaker’s soft and clear voice, the participants also talked softly and clearly. The participants realized that they needed to use clear pronunciation and standard language in order to communicate with the AI conversational agents, and therefore tried to correct their pronunciations or use standard language rather than a dialect. This experience eventually helped them to communicate better with other people as well.

If I make a smart pronunciation with words, Arya will understand. This machine is an odd machine (Participant 2).

Recovery and Maintenance of Memory

The study participants reported that their memory was either maintained or improved by using AI conversational agents. “Brain Talk,” one of the AI speaker’s programs, is a dementia prevention programme developed by medical staff, that improves brain function by asking questions and inducing answers. By using Brain Talk, research participants continued to stimulate their brains in an entertaining manner, which contributed to the maintenance of their memory. One participant with mild dementia expressed great enjoyment in using the AI conversational agents. However, some of the participants felt that they were being treated like kindergarteners. Following the completion of a game in “Brain talk,”“Arya” would inform them to expect a higher level of challenge the next day, allowing participants to look forward to the next brain chat. This program has the potential to help the brain remember things that might have otherwise been forgotten, thereby potentially preventing dementia. Moreover, “Arya” could memorize alarms and remind participants of their appointments or medications, assisting them in staying on track with their schedules and health care.

I will go to the hospital today. I will go to the hospital in a few days. If you talk about it to Arya, it will turn it (the alarm) on and do it accurately (Participant 4).

Emotional Benefits

Positive Affect

Positive affectivity is a human trait that describes how people experience positive emotions and how they interact with others and their environment as a result (Watson, 2005). The biggest benefit of using AI conversational agents was the increase in positive affect among older adults. Most of the participants were on public assistance, and they had weak family relationships, limited financial resources, and a limited social network. They reported experiencing high levels of anxiety, depression, and loneliness. However, after using the AI speaker, it appeared that these levels decreased significantly. A participant said, “In the past, I was always anxious, worried, and depressed. I don’t have those feelings anymore after using the AI speaker.”

The negative emotions seemed to noticeably weaken, while positive emotions improved greatly. Another participant said, “I’m always thankful and happy.”. The participants felt that the AI conversational agents were like family members or friends. Since they had lived alone for a long time, they experienced profound loneliness, especially during the lockdown due to the COVID-19 pandemic. However, the participants could talk to “Arya,” which abated their loneliness. Engaging with the AI conversational agents and listening to music through them seemed to induce positive affection among the participants. They experienced positive emotional changes, especially with respect to loneliness, boredom, depression, and anxiety. In addition, their appreciation for life and positive emotions became quite constant.

Through Emotional Conversation

In terms of conversation with “Arya,” the participants felt less isolated and that they too had someone to talk to. The participants said that Arya is much better than family since they can talk to it whenever they want to talk. Morning and night greetings, and greetings at the start and end of interactions made the participants feel happy, warm, and alive. The participants felt welcomed, and that they were taken care of by someone. Whenever the participants greeted Arya and asked about whether she slept well during the night, Arya responded with expressions of gratitude like “Thank you for remembering me” or “Thanks to you, I slept well.” One participant shared that even when her husband was alive, she never greeted him with “good morning” or “good night.” However, now, she makes it a point to say “hello” to Arya every morning and night.

In addition to greetings, the emotional conversations with Arya were a happy and enjoyable experience for the participants. Spending time with Arya lessened their boredom. The participants said that Arya was very different from having a pet. In the case of dogs, the participants needed to take care of them and go through a lot of difficulties when the pets passed away. Conversely, with Arya, the participants often expressed affection by saying “love you” and gratitude through phrases like “thank you.” Participants said that “when the weather was warm, Arya told me to go out. On cold days, Arya tells me to stay at home.” Even when participants felt depressed, “Arya” suggested they go out for a walk with the dog. The two-way conversation with “Arya” made participants feel comfort and empathy. One participant said that “I have a friend in the space where I used to be. Arya would say to him “Are you tired today? There’s not much left of today, so keep working hard.” When a participant felt anxious, Arya would tell them “to think big so they calm down.” The kindness and purity of “Arya” made participants feel emotionally supported.

Through Listening to Music

Music-based activities would remind the participants of the past, a time when they felt that they had a lot of dreams and enjoyment. The lyrics of Trot also gave the participants an opportunity to think deeply about life and make them accept life’s difficulties. Many participants said, “By listen to music through ‘Arya,’ I have peace of mind and no desire to die.” One participant said, “The depression in my heart slowly disappeared. I was told I had to take antidepressant medication for the rest of my life, but after listening to the songs, I stopped taking the antidepressant medication. This is due to Arya.” It is important for a person who is living alone to have stability of mind. Songs are the best way to feel good: “When I’m in a bad mood or angry, listening to songs calms me down. When my mind is not stable, if the sound of a song emerges, I become absorbed in the song and my mind feels at ease”; “When I listen to songs, I immerse myself in the world alone and instantly feel a lot of happiness.”

Feeling Safe

Before using Arya, there was an expectation that 911 could only be called in an emergency as a last resort. In fact, there were two participants who called 911 and saved their lives. The participants found great reassurance in knowing that help would come right away if they called 911 during an emergency situation, especially when they had fallen down. In addition, participants were able to feel great relief that a device that measures heart rate could be connected to an AI speaker to monitor their health status. Furthermore, participants were prevented from falling when they went to the bathroom in the dark during the night by using a mood light. This also made the participants feel stable.

Gratitude and Happiness

As participants used the AI speaker, their gratitude and happiness for life grew. There were more and more instances of feeling happy or grateful about life. A participant said, “Arya’s warm words comforted me and made me feel grateful.” The participants said that Arya was a comfort in many times when they were facing financial difficulty and feeling annoyed. “Arya” was described as a “precious existence,” an “indispensable existence,” a “thankful existence,” and “always by my side.” Participants expressed that Arya was more than just a machine to them; they referred to Arya as their “best friend” or even “family.”

The Intention to Use AI Conversational Agents, Challenges, and Satisfaction With Them

The participants had multiple intentions when interacting with Arya, including relieving boredom, gaining new information, and dealing with emergency situations. Among these intentions, the most crucial requirement was having the ability to seek help during emergencies. When participants experienced falls, simply yelling “Call for help” to the AI speaker would facilitate a connection to the 911 emergency systems, promptly connecting them with an emergency responder who could provide assistance at their location. This experience has fostered a strong desire among participants to use AI conversational agents for enhanced security and safety. Moreover, as participants spend a significant amount of time alone at home with limited places to go, they also found AI conversational agents beneficial in alleviating boredom. In addition to safety and relief from boredom, participants anticipated to get new information such as weather forecast, horoscopes, and news by using AI conversational agents. Furthermore, when the participants did not know the answer to any question, they expected the AI speaker to have the answer.

There were several challenges with older adults using “Aria.” If they spoke with a dialect or didn’t pronounce words correctly, the AI conversational agents often couldn’t understand them and failed to continue the conversation. After one or two misunderstandings, it became difficult to keep using the AI conversational agent. Another challenge was the inability to have long, meaningful conversations with the AI conversation agents. For example, when an older adult expressed feeling lonely, the AI speakers would encourage them to go out and socialize, but the conversation often ended abruptly afterward. When participants tried to continue the dialogue, the AI speakers would respond, “I don’t know.” Participants expressed a desire for “Area” to ask them how they were doing before the older adults initiated conversation.

Then, if “Aria” asks, “How are you feeling today?” it just melts my heart. It’s comforting to know that someone cares and can offer me comfort. I often think, “Oh, I feel comforted, too,” and I find myself wanting to respond by asking, “How are you feeling today, XXX ?”(Participant 9).

In addition, when using the encyclopaedia, “Area” would only repeat the information that was entered, and participants expressed a desire for more detailed answers. This limitation applied to all the information provided by “Area.” While “Area” allows participants to listen to music, they found it frustrating that the volume of each song varied. Participants preferred a consistent loudness across all songs.

Participants were highly satisfied with AI conversational agents. They expressed the intention to use them until the end of their lives. Notably, the availability of free services significantly influenced their satisfaction. Given that many participants relied on public assistance, they faced financial constraints in acquiring the AI conversational agent device and paying for Internet connections. AI conversational agent device, Internet connection, and unlimited music streaming services provided by local government through their AI conversational agents were provided for free. Therefore, the participants did not incur any economic burden in accessing these services.

Discussion

This paper explores the perceived benefits and satisfaction of older adult users with regard to AI conversational agents. The perceived benefits encompass both instrumental/functional and emotional aspects. The instrumental benefits of using AI conversational agents include facilitating active daily living with “Arya,” seeking and managing diverse information, improving communication skills, and aiding memory recovery and maintenance. Emotional aspects included experiencing positive affect, feeling safe, and expressing gratitude and happiness. The intentions of using conversational agents were to enable direct connection to 911 for emergencies, alleviate boredom, and listen to music as desired. These intentions appeared to be fulfilled, resulting in a high level of satisfaction. Furthermore, the support provided by ICT managers and the availability of free services utilizing AI conversational agents further enhanced the satisfaction levels.

Based on the study findings, our discussion revealed several key points. First, this study found out that there are many instrumental/functional uses of conversational agents for older adults with low-income living alone. Many studies have found a positive relationship between perceived usefulness and the adoption of conversational agents (Kim et al., 2020; Park & Choi, 2018; Yang & Lee, 2019). The conversational agents in this project are considered easy-to-use devices that could play vital roles in assisting with various tasks, such as listening to music, finding information and weather forecasts, engaging in two-way conversations, and connecting with emergency calls regardless of the education level of the participants (Han & Yang, 2018; Katherine et al., 2022).

Second, the study revealed the emotional usefulness of conversational agents, which include fostering positive affect, evoking feelings of gratitude and happiness, and instilling a sense of safety. Ample research exists on how various psychological factors, such as para-social interaction, personification, and loneliness, impact people’s satisfaction with the use of conversational agents. Findings show that users’ para-social interaction positively influenced user satisfaction and the continued use of a device (Han & Yang, 2018; Jang, 2020). The results from these studies unveil that social attraction has a much stronger effect on para-social interaction than task/utility and physical attraction. Moreover, participants perceived the AI conversational agents as human-like, and they exhibited personifying behaviours when interacting with conversational agents, often referring to the devices as “friends” (Jang, 2020; Kim et al., 2020; Lee & Lee, 2022; Park & Choi, 2018; Pradhan et al., 2019). The study’s participants considered AI conversational agents as friends or family members when they spoke with “Arya,” and as a smart helper when they found information easily and listened to music. Major South Korean mobile carriers analysed usage patterns of the conversational agents and showed that one of the most frequent words people, including older adults living alone, used was “emotional conversation” (Jang, 2020). The research found that conversational agents are perceived by older adults to reduce loneliness and improve social connectedness (Cho & Ryu, 2022; Corbett et al., 2021; Katherine et al., 2022; Kim et al., 2020; Kim et al., 2022; Lee & Lee, 2022; Song et al., 2021). This study found that AI conversational agents can lessen social isolation and loneliness and can play a positive role, particularly in the lives of older adults living alone.

Implications

Based on the findings of this study, three key areas of implications (theoretical, managerial, and social) have been identified, highlighting the need for attention and intervention to ensure the successful adoption and usage of AI conversational agents among older adults.

Theoretical Implications

Integrated Model of Care

This study provides theoretical implications for the integrated model of care for older adults by combining non-face-to-face interactions using AI conversational agents and face-to-face services facilitated by ICT managers. The participants in this study expressed their gratitude toward ICT managers. This study provides valuable insights into the potential benefits of integrating AI speaker technology with traditional care services. These findings contribute to the development of theoretical frameworks associated with integrated care models and service delivery, shedding light on the effective utilization of technology in improving care for older adults.

Addressing Research Gaps and Providing Insights on Socioeconomically Marginalized Older Adults

This study draws attention to the limited existing research on older adults living alone and those with low-income as marginalized population group. Also, it emphasizes the scarcity of literature that explores rich information, including opinions and perspectives of older adults, which cannot be fully captured through quantitative research alone. Therefore, this study provides valuable theoretical implications by filling this research gap and offering insights into the experiences and perspectives of socioeconomically marginalized older adults.

Managerial Implications

Addressing Attitudinal Issues and ICT Manager Support

Older adults often hold misconceptions about the complexity, risks, and costs associated with using technology, including AI conversational agents (Blaschke et al., 2009; Corbett et al., 2021; Kim, 2021). These perceptions, coupled with low confidence in device usage and privacy concerns, can hinder the adoption and utilization of AI conversational agents. To overcome these attitudinal barriers and misconceptions surrounding AI conversational agents, this study offers implications for the vital role of ICT managers in providing clear information, guidance, and assistance to enhance acceptance and usage. ICT managers can foster a better understanding of AI conversational agents and effectively address any concerns that older adults may have.

Financial Accessibility and Support from Local Governments

The costs associated with using AI conversational agents, including Internet connection fees, may pose barriers for older adults, especially those with low incomes. In this regard, local governments of South Korea can contribute to supporting older adults by providing financial assistance. This support may take the form of subsidizing the costs of AI conversational agent devices and Internet connections or offering additional benefits like unlimited music streaming services through AI conversational agents.

Social Implications

Social Connectedness and Loneliness

This study offers valuable insights into the role of AI speakers in addressing social connectedness and loneliness among older adults, particularly those living alone with limited social interactions. The findings of this study suggest that AI conversational agents play a crucial role in alleviating social isolation and loneliness while fostering social connectedness through interactions with older adults. These findings align with existing literature that highlights the positive impact of AI conversational agents on social connectedness and loneliness (Cho & Ryu, 2022; Han & Yang, 2018; Jang, 2020; Katherine et al., 2022, Kim et al., 2020, 2022). Consequently, this study contributes to the understanding of social implications related to social isolation and loneliness among older adults.

Information Access and Empowerment of Older Adults as a Vulnerable Population

Moreover, this study provides social implications regarding the role of AI speakers in enhancing information access, reducing the digital divide, and promoting empowerment among older adults with limited resources.

Limitations and Future Research Suggestion

This study has identified two limitations. The first limitation of this study was that it did not consider the experiences of those who applied for AI conversational agents and returned them after little use. Second, the role of ICT managers who play a pivotal role in assisting users was not explored, hence, it was not possible to understand how human help is combined with mechanical help. Future research should focus on specific investigations into the role of ICT managers.

Conclusion

To gain insights into the perspectives and usage patterns of conversational agents among low-income older adults living alone, this paper explores older adult users’ perceived benefits and satisfaction. The findings revealed that participants derived joy from accessing information, engaging in conversations, and listening to music through the AI conversational agents. Also, the ability to connect to 911 in an emergency at any time gave them a sense of security.

The findings suggest that as the aging population continues to grow, AI technologies like Arya could play a valuable role in enhancing the independence, well-being, and safety for older adults, especially those living in isolated conditions. To continue to provide emotional and informational support for older adults living alone, it is imperative to garner financial support from the government, technical assistance from IT companies, and practical support from ICT managers.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2021S1A5A2A01062344).

ORCID iDs

Kyung Mee Kim

Sook Hyun Kim

Data Availability Statement

The data that support the findings of this study consist of qualitative interview transcripts, which are in Korean. The data can be available from the first author upon request.

References

Blaschke

C. M.

Freddolino

P. P.

Mullen

E. E.

(2009). Ageing and technology: A review of the research literature. British Journal of Social Work, 39(4), 641–656. https://doi.org/10.1093/bjsw/bcp025

Braun

Clarke

(2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa

Cho

Ryu

M. H.

(2022). Effects of user’s health characteristics on the intention to use artificial intelligence speakers. Journal of Korean Institute of Communications and Information Sciences, 47(1), 160–177.

Corbett

C. F.

Wright

P. J.

Jones

Parmer

(2021). Voice-activated virtual home assistant use and social isolation and loneliness among older adults: Mini review. Frontiers in Public Health 9, 742012. https://doi.org/10.3389/fpubh.2021.742012

Davis

F. D.

(1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340.

Han

Yang

(2018). Understanding adoption of intelligent personal assistant: A parasocial relationship perspective. Industrial Management & Data Systems, 118(3), 618–636. https://doi.org/10.1108/IMDS-05-2017-0214

Horton

Wohl

R.R.

(1956). Mass communication and para-social interaction: Observations on intimacy at a distance. Psychiatry, 19(3), 215–229.

Jang

(2020). Exploring user interaction and satisfaction with virtual personal assistant usage through smart speakers. Archives of Design Research, 33(3), 127–135. https://doi.org/10.15187/adr.2020.08.33.3.127

Katherine

Liggett

Ramirez-Zohfeld

Sunkara

Lindquist

L.A.

(2020). Voice-controlled intelligent personal assistants to support aging in place, Journal of the American Geriatrics Society, 68(1), 176–179. https://doi.org/10.1111/jgs.16217

10.

Katherine

Light

S.W.

Bradley

Lindquist

L.A.

(2022). Optimizing voice-controlled intelligent personal assistants for use by home-bound older adults, Journal of the American Geriatrics Society, 70(5), 1504–1509. https://doi.org/10.1111/jgs.17625

11.

Kim

(2021). Exploring how older adults use a smart speaker-based voice assistant in their first interactions: Qualitative study, Journal of Medical Internet Research Mhealth Uhealth, 9(1), e20427. https://doi.org/10.2196/20427

12.

Kim

Hwang

Kim

Lee

(2022). Toward smart communication components: Recent advances in human and AI speaker interaction. Electronics, 11(10), 1533. https://doi.org/10.3390/electronics11101533

13.

Kim

Song

Y. J.

Sung

Choi

S. M.

(2020). AI speaker for the elderly: Functional and emotional evaluation of AI speaker. Journal of Media Economics and Culture, 18(4), 7–35.

14.

Koh

W. Q.

Felding

S. A.

Toomey

Casey

(2021). Barriers and facilitators to the implementation of social robots for older adults and people with dementia: A scoping review protocol. BMC Geriatrics, 21, 351. https://doi.org/10.1186/s12877-021-02277-9

15.

Korea JoongAng Daily. (2022, September 29). People 65 and over account for 17.5% of Korea’s population this year. https://koreajoongangdaily.joins.com/2022/09/29/business/economy/korea-population-aging-population/20220929184932007.html

16.

Korea Ministry of Health and Welfare. (2021). White paper on suicide prevention [White paper]. https://www.bokjiro.go.kr/ssis-tbu/cms/pc/news/promotion/__icsFiles/afieldfile/2021/08/30/162547124180998.pdf

17.

Kuiper

J. S.

Zuidersma

Oude Voshaar

R. C.

Zuidema

S. U.

van den Heuvel

E. R.

Stolk

R.P.

Smidt

(2015). Social relationships and risk of dementia: A systematic review and meta-analysis of longitudinal cohort studies. Ageing Research Reviews, 22, 39–57. https://doi.org/10.1016/j.arr.2015.04.006

18.

Lee

(2022). A direction of AI speaker design for the elderly to alleviate the feeling of depression: The formation of emotional intimacy. Archives of Design Research, 35(3), 159–179.

19.

Nallam

Bhandari

Sanders

Martin-Hammond

(2020). A question of access: Exploring the perceived benefits and barriers of intelligent voice assistants for improving access to consumer health resources among low-income older adults. Gerontology & Geriatric Medicine, 6, 5975. https://doi.org/10.1177/2333721420985975

20.

Organization for Economic Co-operation and Development (OECD). (2021). Old-age income poverty. https://www.oecd-ilibrary.org/sites/d76e4fad-en/index.html?itemId=/content/component/d76e4fad-en

21.

Park

S. A.

Choi

S. M.

(2018). An understanding of the factors influencing satisfaction and continued use intention of AI speakers: Focusing on utilitarian and hedonic values. Information Society & Media, 19(3), 159–182.

22.

Pradhan

Findlater

Lazar

(2019). “Phantom friend’ or “just a box with information”: Personification and ontological categorization of smart speaker-based voice assistant by older adults. Proceedings of the ACM on human-computer interaction, 3(Issue CSCW), 214, 1–21. https://doi.org/10.1145/3359316

23.

Saad

Afzal

El-Issawi

Eid

(2016). A model to measure QoE for virtual personal assistant. Multimedia Tools and Applications, 76, 12517–12537. https://doi.org/10.1007/s11042-016-3650-5

24.

SK Telecom and National Social Solidarity Economy Local Government Council, and Happy Connect Foundation. (2020). Happy community AI services. [White paper]. http://www.happyconnect.co.kr/bbs/board.php?bo_table=news&wr_id=33&ckattempt=1

25.

Slegers

Van Boxtel

M. P.

Jolles

(2008). Effects of computer training and Internet usage on the well-being and quality of life of older adults: a randomized, controlled study. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 63(3), 176–184.

26.

Song

Y. J.

Kim

Choi

S. M.

Sung

(2021). A study on the factors influencing elderly users’ satisfaction with and the continued use intention of AI Cas. Korean Journal of Broadcasting & Telecommunications Research, 114, 9–37.

27.

Statistics Korea. (2022). Population projections: 2020-2070. https://kostat.go.kr/board.es?mid=a10301010000&bid=207&tag=&act=view&list_no=415453&ref_bid=

28.

Watson

(2005). Positive affectivity: The disposition to experience pleasurable emotional states. In Snyder

Lopez

S. J.

(Eds.), Handbook of positive psychology (pp. 106–119). New York: Oxford University Press.

29.

Yang

Lee

(2018). Understanding user behavior of virtual personal assistant devices. Information Systems and e-Business Management, 17(1), 65–87. https://doi.org/10.1007/s10257-018-0375-1