1. Introduction
In recent years, South Korea has faced a growing social challenge known as solitary death (godoksa), where individuals—often elderly—die alone and remain undiscovered for extended periods [1]. This phenomenon has drawn increasing attention in light of South Korea’s rapidly aging population, accelerated urbanization, and the erosion of traditional family structures. The rise in solitary deaths has raised alarm over the well-being of vulnerable groups, including older adults living alone, individuals with limited social networks, and those facing economic hardship [1].
Solitary death is not merely a demographic issue but also a reflection of broader societal transformations. Traditional social safety nets, such as intergenerational family support and neighborhood-based community care, have weakened in recent decades [2]. Furthermore, cultural stigma associated with seeking help often deters those at risk from accessing support services [2]. These interrelated factors highlight the urgency of developing effective and culturally responsive interventions to prevent solitary deaths and to ensure timely assistance for those in need.
In response to this pressing issue, major South Korean technology companies have developed and deployed Artificial Intelligence (AI)-based eldercare services aimed at reducing the risk of solitary death. Notably, Naver’s CLOVA Carecall, KT’s AI Care Speaker, and SK Telecom’s NUGU Carecall have emerged as prominent AI-driven platforms [3]. These services monitor daily routines and psychological well-being of at-risk individuals, enabling early intervention when anomalies in behavior or communication are detected. For example, automated check-in calls analyze interaction patterns and notify caregivers or authorities when signs of distress or abnormal inactivity are observed.
This paper examines the effectiveness and limitations of these AI-based eldercare services in reducing solitary death risk among older adults in South Korea. The study focuses on their sociotechnical integration, ethical challenges, and policy implications within the South Korean context.
2. Overview of AI-Based Care Services in South Korea
In response to the growing concern about solitary deaths, particularly among older adults living alone, various AI-based care services have been developed and implemented in South Korea. This section introduces and analyzes three representative platforms—Naver’s CLOVA Carecall, KT’s AI Care Speaker, and SK Telecom’s NUGU Carecall—focusing on their operational models, target populations, and reported outcomes.
2.1 Naver’s CLOVA Carecall
CLOVA Carecall, developed by Naver Corporation, is a voice-based AI call service designed to monitor and support older adults living alone [4]. Launched in partnership with local governments, the system conducts automated daily check-in calls using natural language processing to assess users’ emotional states and detect signs of distress or abnormal behavior. If concerning patterns are detected—such as unresponsiveness or expressions of negative emotions—the system alerts local welfare officials for follow-up [5].
Empirical data from pilot programs in cities such as Yongin and Seongnam suggest a reduction in response time for at-risk individuals. According to municipal reports, welfare workers were able to intervene promptly in several instances where the AI system detected irregular speech patterns or prolonged silence. As of 2023, CLOVA Carecall had been adopted in over 100 municipalities, with continued expansion encouraged by the Ministry of Health and Welfare [4].
2.2 KT’s AI Care Speaker
The AI Care Speaker is a smart speaker platform distributed primarily to older adults living alone [6]. The device offers various features, including voice-activated emergency alerts, daily health check-ins, weather updates, and audio-based mental stimulation activities such as quizzes and music.
A nationwide initiative launched in 2021 saw over 30,000 units distributed in partnership with local governments. Preliminary assessments indicate that users reported increased feelings of safety and reduced subjective loneliness, particularly in rural areas with limited access to in-person care services. The device is also linked to local emergency response systems, allowing for immediate action when emergency keywords (e.g., “help,” “pain”) are detected [7].
Despite these benefits, several usability challenges have been reported. Some users—especially the oldest-old—experienced difficulty in interacting with the device due to unfamiliarity with voice-activated technology or hearing impairments. These limitations underscore the importance of combining technological interventions with human instruction and support [7].
2.3 SK Telecom’s NUGU Carecall
NUGU Carecall, operated by SK Telecom, provides AI-powered phone calls aimed at regular emotional check-ins with older adults [8]. Unlike CLOVA, NUGU focuses less on detecting emotional tone and more on ensuring the recipient is maintaining routine behaviors such as eating, taking medication, and sleeping regularly. The AI agent follows a structured dialogue model and records users’ responses, which are then monitored by local case managers.
In a collaboration with the Seoul Metropolitan Government and other regional partners, NUGU Carecall was deployed to over 20,000 older adults by the end of 2022. Feedback from case managers indicates that the service improved efficiency in identifying behavioral anomalies, particularly among older adults with mild cognitive impairment or early dementia. Instances of missed meals or medication non-compliance were more frequently detected and addressed [9].
However, the service’s reliance on phone communication (without a screen or physical interface) also limits its capacity for rich interaction, and may miss subtler emotional cues. User satisfaction surveys show moderate to high acceptance, but also a strong preference for human follow-up, reinforcing the view that such services function best as supplementary tools [9].
Table 1
Table 1. Features of AI-Based Care Services
| Feature |
Naver’s
CLOVA Carecall |
KT’s
AI Care Speaker |
SK Telecom’s
NUGU Carecall |
|---|---|---|---|
| Service Type | Voice-based automated check-in calls | Smart speaker with voice-activated features | AI-powered phone calls with structured dialogue |
| Primary Focus | Emotional state monitoring & distress detection | Emotional support, safety alerts, mental stimulation | Monitoring daily routines: meals, medication, sleep |
| Deployment Scale | Adopted in 100+ municipalities (e.g., Yongin, Seongnam) | 30,000+ units nationwide (especially rural areas) | 20,000+ older adults in Seoul and regional partners |
| Key Technologies | Natural language processing (NLP), AI call analysis | Voice recognition, emergency keyword detection | Structured dialogue system, behavioral anomaly detection |
| Reported Benefits | Faster welfare response; early distress detection | Reduced loneliness; increased peace of mind | Improved detection of behavioral irregularities; clinical referrals |
| Limitations | May miss subtler emotional cues; reliance on phone calls | Usability challenges for oldest-old; hearing impairments | Lack of rich interaction; preference for human follow-up |
| User Population | Older adults living alone in urban/suburban areas | Older adults living alone, especially rural | Older adults, including those with mild cognitive impairment |
| Follow-up | Alerts local welfare officials for intervention | Linked to emergency response systems | Responses monitored by local case managers |
3. Effectiveness and Challenges of AI-Based Interventions
This section evaluates the effectiveness of AI-based interventions in South Korea and identifies the primary challenges encountered during their implementation. Drawing on data from pilot programs and municipal records between 2021 and 2023, we assess three platforms— Naver’s CLOVA Carecall, KT’s AI Care Speaker, and SK Telecom’s NUGU Carecall —in terms of risk detection, user engagement, system integration, and operational limitations.
3.1 Effectiveness in Reducing Solitary Death Risk
The CLOVA Carecall pilot project demonstrated promising results in preventing and managing solitary deaths among older adults [10]. This AI-driven voice call service regularly monitors the well-being of vulnerable individuals, enabling early detection of abnormal health or behavioral signs. During the monitoring period, CLOVA Carecall facilitated timely follow-up actions and clinical referrals, thereby strengthening the safety net for cognitively and physically vulnerable elderly populations. The evaluation underscores CLOVA Carecall’s potential as an effective community-based tool for solitary death prevention through continuous remote monitoring and rapid intervention.
AI speaker-based care services have also shown significant effectiveness in reducing depressive symptoms among elderly users [11]. Regular interaction with AI speakers not only alleviates feelings of loneliness but also enhances emotional well-being by providing timely reminders, social engagement, and mental stimulation. These services offer accessible, continuous support, contributing to improved mental health outcomes and a better quality of life for older adults living alone or with limited social contact.
NUGU Carecall employs a structured dialogue model focusing on daily behaviors such as eating, sleeping, and medication adherence. SK Telecom [12] reported that NUGU Carecall effectively reduces risks related to solitary deaths by detecting abnormal responses promptly and facilitating swift intervention. By ensuring continuous remote monitoring, NUGU Carecall helps improve emotional stability and health management among vulnerable elderly populations, thereby supporting community-based solitary death prevention efforts.
3.2 Challenges in Implementation and Accessibility
Older adults with visual, auditory, or cognitive impairments experienced significant difficulty in engaging with AI services, particularly smart speakers. In several municipalities, users required repeated assistance from family members or community staff to operate devices. This digital divide constrained service accessibility, especially among the oldest-old (aged 80+), highlighting the need for inclusive user interface design and ongoing digital literacy education [13].
While AI systems successfully identified risk signals, the effectiveness of follow-up interventions depended heavily on local welfare capacity. Municipalities with established case management protocols were able to respond quickly to AI-generated alerts, while others lacked sufficient staff to ensure timely interventions. These disparities suggest that AI systems are not self-sufficient; they function optimally only when integrated with robust human support networks.
Some recipients and their guardians expressed concern over the collection and use of sensitive personal data, particularly in voice-based systems. Issues included lack of clarity regarding consent procedures, data retention policies, and the potential for misuse of AI-collected information. Ethical frameworks for data governance and transparent communication remain underdeveloped, posing a barrier to broader public acceptance.
4. Policy and Practical Implications
The empirical analysis of AI-based eldercare interventions in South Korea highlights both the transformative potential and critical limitations of such technologies in addressing the phenomenon of solitary death. Building on observed outcomes and implementation challenges, this section outlines key policy considerations and practical strategies necessary to enhance the sustainability, ethical integrity, and social effectiveness of AI-driven care models.
4.1 Toward a Hybrid Welfare Infrastructure
The findings underscore the necessity of positioning AI technologies as complementary tools, rather than replacements, within the broader eldercare ecosystem. AI interventions such as CLOVA Carecall and NUGU Carecall effectively supported early detection and monitoring; however, their impact was maximized only when embedded within local welfare systems capable of providing follow-up services. Policymakers should adopt a hybrid model that integrates AI with human-delivered care, ensuring that digital tools are aligned with regional capacities, case management systems, and social work expertise.
To support this, national policy frameworks must prioritize the institutionalization of AI-assisted care as a formal component of community welfare planning. This includes developing standardized protocols for AI alert responses, training caseworkers in technology-enabled monitoring, and expanding funding mechanisms for municipalities with limited digital infrastructure.
4.2 Ethical Governance and User Protection
The rapid deployment of AI in public welfare has outpaced the development of adequate data protection policies, particularly in relation to older adults who may lack capacity to provide informed consent. Concerns around surveillance, algorithmic decision-making, and voice data storage were frequently raised by both users and care providers during the pilot implementations [14].
To address this, the government should implement ethically grounded regulatory standards that ensure transparency in AI operations, user autonomy in data sharing, and accountability in automated risk assessments. This may involve the creation of an independent oversight body to monitor AI use in social services and the introduction of AI-specific consent procedures adapted for cognitively vulnerable populations.
4.3 Digital Inclusion and User-Centered Design
Unequal access to AI-based care remains a central challenge, particularly for older adults with limited digital literacy or sensory impairments. For AI interventions to be equitable, national strategies must include targeted digital inclusion initiatives. These may involve community-based digital literacy education tailored for older adults, the development of universal design standards for AI interfaces that accommodate physical and cognitive limitations, and increased provision of technical support services, including home visits by trained personnel to assist in device use and troubleshooting.
Digital inclusion should be viewed not only as a technological issue but also as a social determinant of health and aging.
4.4 Sustainability and Long-Term Planning
While the early results of AI-based eldercare programs are promising, their long-term sustainability remains uncertain. Most pilot projects have been funded through short-term public-private partnerships or emergency welfare grants (e.g., during the COVID-19 pandemic). To ensure continuity and scalability, the central government must incorporate AI care services into long-term national aging and welfare plans.
This includes allocating stable budgetary resources for AI service maintenance and upgrades, promoting inter-ministerial coordination between health, ICT, and welfare sectors, and supporting research and development in gerontechnology with a focus on high-risk groups such as those experiencing social isolation, poverty, or disability.
4.5 International Relevance and Future Research
South Korea’s proactive adoption of AI to address solitary death presents a valuable model for other aging societies facing similar demographic and social challenges. However, further research is needed to evaluate long-term outcomes of AI interventions on mortality, mental health, and quality of life, identify cross-cultural differences in technology acceptance among older adults, and explore the integration of AI with other digital health innovations such as remote diagnostics or wearable monitoring systems.
Comparative studies with countries implementing parallel interventions (e.g., Japan’s robotics in eldercare or the UK’s digital companionship programs) would offer useful insights into global best practices.
5. Conclusion
As South Korea confronts the complex social issue of solitary death, especially among its rapidly aging population, AI-based interventions have emerged as innovative tools within the national and municipal welfare landscape. This study analyzed three representative platforms—Naver’s CLOVA Carecall, KT’s AI Care Speaker, and SK Telecom’s NUGU Carecall—offering insights into their practical effectiveness, operational limitations, and broader implications for policy and practice.
The findings suggest that AI technologies can play a meaningful role in reducing risk factors associated with solitary death by providing consistent monitoring, emotional support, and timely alerts for irregularities in older adults’ daily routines. In particular, these platforms have demonstrated potential for enhancing social connectedness and supplementing human welfare services, especially in resource-limited settings.
However, the implementation of AI services is not without critical challenges. Issues of digital accessibility, technological literacy, data privacy, and ethical governance remain central to the discourse. Moreover, the effectiveness of these systems is contingent upon their integration with human-centered care infrastructures. AI alone cannot replace the social and relational dimensions of eldercare; rather, its value lies in augmenting human capacity and enabling more responsive and preventative service models.
To fully realize the benefits of AI in aging societies, a hybrid care paradigm must be established—one that blends technological innovation with ethical responsibility, inclusivity, and sustainable policy support. South Korea’s experience offers a valuable case study for other nations grappling with similar demographic trends, providing a foundation for developing globally relevant, compassionate, and intelligent eldercare systems.
Future research should continue to assess long-term outcomes of AI interventions, explore their adaptability across diverse populations, and examine the institutional conditions that facilitate or hinder their success. As the line between welfare and technology continues to blur, there is an urgent need for interdisciplinary collaboration to ensure that AI advances contribute meaningfully to human dignity, well-being, and the right to age safely and socially connected.
CONFLICTS OF INTEREST
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
References
H. Y. Kim and T. H. Ha, “With Corona Era, exploring policy measures to prevent non-face-to-face lonely deaths – Focusing on Daegu Metropolitan City’s AI and IOT cases of lonely death prevention,” Journal of Digital Convergence, vol. 21, no. 3, pp. 49-62, Dec. 2023, doi: https://doi.org/10.14400/JDC.2023.21.3.049.
J. W. Kim, Y. J. Song, Y. J. Sung, and S. M. Choi, “AI speaker for the elderly: functional and emotional evaluation of AI speaker,” Journal of Media, Economics, & Culture, vol. 18, no. 4, pp. 7-35, Nov. 2020, doi: http://dx.doi.org/10.21328/JMEC.2020.11.18.4.7.
M. Oh, S. Kim, N. Y. Park, S. An, and Y. Cho, A study on the applicability and prospect of digital technology in the health and social welfare, Korea Institute for Health and Social Affairs, Sejong, 2022, doi: https://dx.doi.org/10.23060/kihasa.a.2022.37.
S. Y. Choi, “A study on human-centered IT utilization in caring for elderly people who live alone,” Journal of Digital Convergence, vol. 20, no. 2, pp.455-462, Feb. 2022, doi: https://doi.org/10.14400/JDC.2022.20.2.455.