Voice assistants in private households: a conceptual framework for future research in an interdisciplinary field

The paper addresses how to make sense of a rapidly growing and fragmented body of research on voice assistants (VAs) in private households, and how interdisciplinary insights can guide future use cases and business opportunities. Positioned within the broader context of digitization as a driver of economic development, the study notes that VAs—enabled by speech-driven interaction and AI—have seen rapid consumer adoption via smart speakers (e.g., Amazon Echo, Apple HomePod, Google Home). Adoption figures indicate strong market traction (e.g., 15.4% of US and 5.9% of German populations owned an Amazon Echo by 2018; smart homes in Europe and North America reached 105 million in 2021). Yet, concerns around privacy, passive listening, and data security remain prominent and may hinder adoption of complex VA applications. The research question focuses on identifying, organizing, and integrating dispersed scholarly insights across Computer Science (CS), Social Science (SS), and Business & Management Science (BMS) to conceptualize linkages and common themes, inform future research directions, and illuminate viable business opportunities in private-home VA use. Drawing on the IDEO innovation lenses of feasibility (technical/legal), viability (economic), and desirability (user and societal acceptance), the paper argues that an interdisciplinary approach is essential to align technological development with social acceptance and business models.

The paper synthesizes literature on business opportunities and adoption considerations for VAs in private households. Positives include projected efficiency gains (voice control of household devices, smart home orchestration), health-related applications (patient communication, self-care), and high reported usage for home management (54% of surveyed US smart speaker users). Envisioned future uses span broader home controls and smart services. However, negatives are substantial: surveillance-capitalism concerns, privacy/security risks, passive listening worries (52% data security and 41% passive listening concerns in a Microsoft/Bing survey), unclear legal frameworks, and questions around failure modes and rights of autonomous systems. Despite technological feasibility of many use cases, current household use remains largely for simple tasks (music, quick facts, directions, news/weather). The review highlights that VA business opportunities differ from in-firm technologies (e.g., RFID, IoT process reengineering) because benefits depend on widespread consumer adoption of hardware/services. Consequently, interdisciplinary integration of CS, SS, and BMS insights is necessary to craft secure, user-valued offerings and sustainable business models, and to navigate ecosystem-level challenges (regulation, insurance, real estate, partnerships).

Study design: systematic literature review combining automated bibliometric analysis with qualitative content analysis to structure dispersed research across CS, SS, and BMS and derive integrated research avenues. - Data source: Scopus (coverage comparable to Web of Science). Articles published in English, in CS, SS, and BMS, before May 2020. - Search strategy: Keywords included “voice assistant” and synonyms (“virtual assistant,” “intelligent personal assistant,” “voice-activated personal assistant,” “conversational agent,” “SIRI,” “Alexa,” “Google Assistant,” “Bixby,” “smart loudspeaker,” “Echo,” “smart speaker”) combined with “home/house/household.” Search targeted title, abstract, and keywords. - Screening framework: PRISMA guidelines applied. Initial results n=824 (CS 428; SS 356; BMS 40). After relevance screening and data cleaning, a working dataset was formed (figure indicates n=267). Subsequent qualitative content analysis excluded 60 items (5 duplicates; 55 not about VAs in private households), yielding a final sample of 207 articles. Final disciplinary assignment (Table 4) resulted in 147 CS, 47 SS, and 13 BMS papers across clusters. - Bibliometric analysis: Conducted separately per discipline using VOSviewer. Generated keyword co-occurrence networks to surface themes/clusters. A thesaurus grouped synonyms and standardized terms; some unrelated keywords were removed to maintain focus. Scopus Subject Areas guided discipline-specific sets; some papers were cross-assigned when appropriate. - Qualitative content analysis: Three researchers independently and then collaboratively assigned each of the 267 screened articles to nine thematic clusters derived from the bibliometric maps, following Krippendorff’s methodology. Investigator triangulation reduced subjectivity. The team refined cluster membership, removed non-relevant items, wrote cluster summaries, and consolidated clusters into four interdisciplinary research streams through iterative qualitative synthesis (independent conceptualization followed by team discussion and agreement on stream definitions and allocations). - Descriptive insights: Top publishing countries included the United States (CS: 66; SS: 19), India (CS: 21), United Kingdom (CS: 18; SS: 5), Germany (CS: 15; SS: 4), among others, indicating a geographically diverse discourse with strong US representation. - Outputs: Nine thematic clusters and four integrated research streams; propositions and a conceptual framework for interdisciplinary future research; identification of business opportunity areas and associated disciplinary challenges.

- The field is fragmented across CS, SS, and BMS with limited cross-fertilization despite technological progress in VAs; privacy, security, and social acceptance issues constrain adoption of complex applications in homes. - Nine thematic clusters identified: 1) Smart devices: VA-controlled IoT/smart home orchestration; user choices/concerns (privacy; impacts on children); technical limitations (language options, transmission range, security, training, context awareness). 2) HCI and UX: User experience challenges (unmet expectations, language issues), trust and acceptance drivers (usefulness, access to data, perceived advantages), and anthropomorphism/humanized VAs, including older adults’ preferences. 3) Privacy and technology adoption: Predominantly CS security/privacy solutions; across disciplines, perceived privacy risks reduce adoption; calls for standardized data frameworks and legal protections. 4) VA marketing strategies: BMS work on advertising, measuring satisfaction, and highlighting security/usability improvements. 5) Technical challenges in VA application development: Speech robustness, watermarking to prevent cross-device misdetection, knowledge graphs, cross-lingual dialogue, fog computing, and dynamic service integration; parallel SS work on affective computing not well-linked to CS. 6) Potential future applications and developments: Prototypes and prospects in therapy/medical care, content creation, and dialogue models; joint CS–SS focus on language/dialogue, near-future scenarios (libraries, assisted living, emergency support), and differences between home vs. public settings. 7) Efficiency increase by VA use: Efficiency gains in home automation, assistive interfaces, activity assistance, elderly care; BMS on efficiency awareness; limited SS contributions. 8) VAs providing legal evidence: Digital forensics potential, data types, tools, and approaches relevant to court cases. 9) VAs supporting assisted living: Technical solutions for daily tasks, companionship, stress management, distress recognition; usability and acceptance for elderly; CS–SS interest in self-determined living and home therapy. - Four integrated research streams: 1) Conceptual foundations (theory of VA design, anthropomorphism, adoption, security/privacy concepts). 2) Systemic challenges, enabling technologies, implementation (security/UX in multi-user contexts, customization, ML, device setup, legal regulations/GDPR/accountability). 3) Efficiency (marketing strategies, raising awareness of benefits, addressing low perceived utility). 4) Applications and use cases (user perspectives, prototypes, overcoming limitations, medical/fitness/assisted living, IoT forensics). - Cross-discipline contributions: CS dominates technical advancements and systemic solutions; SS contributes theoretical foundations and user perceptions; BMS focuses on efficiency and marketing, with gaps in security and assisted living nuances. - Market and adoption data points: • By 2018, 15.4% (US) and 5.9% (Germany) owned Amazon Echo; private household purchases grew 116% in Q3 2018 vs 2017. • 105 million smart homes in Europe and North America in 2021. • 54% of surveyed US users use smart speakers to manage homes; top concerns: data security (52%) and passive listening (41%). - Business opportunities and maturity: • Smart home systems: high technical maturity; high SS concerns (privacy); medium BMS challenge (privacy-by-design business models). • Assisted living/medical home therapy: significant CS (affective computing), SS (emotional support needs), and BMS (public–private models) challenges. • Digital forensics: rising importance with high CS/SS/BMS challenges (tools, legal frameworks, business models). - Overarching insight: Advancing adoption requires interdisciplinary integration to address security/privacy, define perceived-safe environments, articulate benefits beyond efficiency, and model viable business cases.

The findings demonstrate that while VAs are technologically feasible and increasingly present in homes, user trust and perceived safety remain pivotal barriers to broader adoption of complex applications. Fragmentation across CS, SS, and BMS inhibits translation of technical advances into desirable, viable, and legally compliant solutions for private households. - Addressing the research question, the consolidation into nine clusters and four streams clarifies complementary roles: CS to deliver secure, integrated, explainable systems; SS to explicate user needs, social norms, and legal expectations; BMS to articulate value propositions beyond efficiency and to design ecosystem-aware business models. - The study advances propositions for future work: • Proposition 1: Interdisciplinary integration across CS, SS, BMS is necessary to increase adoption of complex VA applications in homes. • Proposition 1.1: Integrate SS insights on perception/security with CS solutions and BMS business modeling to realize safe medical-care use cases. • Proposition 1.2: Advance systemic integration and security (CS) alongside regulation to foster perceived safety and smart-home efficiency. • Proposition 1.3: Use user research and near-future scenarios to develop/test BMS/SS-informed business cases beyond efficiency assumptions. • Proposition 2: Interdisciplinary efforts are needed to overcome ecosystem-level adoption challenges (regulation, insurance, real estate, partnerships). - Significance: The proposed framework (mapping streams to propositions) guides scholars and managers toward aligned, ecosystem-centric strategies for VA services. It underscores the need for transparent data frameworks, robust security, legal clarity, and compelling user benefits (e.g., health, autonomy, well-being) to build trust and drive uptake.

This review consolidates dispersed literature on VAs in private households into nine thematic clusters and four interdisciplinary research streams, revealing limited cross-fertilization between CS, SS, and BMS. It proposes a conceptual framework and propositions to guide integrated future research, emphasizing security/privacy resolutions, legal/regulatory clarity, and user-centric value beyond efficiency. Business opportunities are most promising in smart home systems, assisted living/medical home therapy, and digital forensics, each with distinct maturity and challenge profiles. The paper anticipates VA evolution beyond audio-only devices toward multimodal, embodied assistants (e.g., with screens/cameras, social robots), requiring technical integration and nuanced studies of user perception. For managers, success depends on designing trustworthy, transparent ecosystems and partnerships, not solely on efficiency gains. Future research should expand database coverage, domains, and methods (e.g., co-citation/bibliographic coupling), and investigate post-COVID-19 shifts in home VA use.

- Method subjectivity: Despite investigator triangulation, qualitative synthesis and cluster/stream formation may introduce author subjectivity. - Database scope: Only Scopus-indexed articles were included; relevant work in EBSCO, Web of Science, Google Scholar may be missing. - Domain and time bounds: Focused on CS, SS, and BMS up to May 2020; evolving contexts (e.g., COVID-19 impacts) were not analyzed. - Data cleaning constraints: Initial keyword searches returned irrelevant items (e.g., “echo” as an acoustic term), requiring manual cleaning. - Future methodological enhancements: Co-citation and bibliographic coupling analyses across authors/institutions/countries could complement findings.