Generative AI and the human touch: unpacking well-being in SME services

In a small coffee shop nestled on a small street in Ho Chi Minh City, the owner begins her morning by checking her phone, not for messages, but for insights. Her newly integrated generative artificial intelligence (GenAI) assistant has analyzed customer reviews overnight, predicted the preferences of returning customers and even suggested sustainable packaging solutions based on market trends. When a loyal customer walks in later that morning, the owner greets her by name and already knows she prefers oat milk and is interested in eco-friendly products. The customer smiles, not just because the coffee is perfect, but because she feels seen, understood and part of something more meaningful. This is not just convenience. This is value-in-use in action which is generated by hybrid intelligence which include the human and GenAI. It is the real-time, human-centered benefit that customers experience when artificial intelligence (AI) works quietly in the background to personalize services, respect their values and make their lives just a bit easier. A small-entrepreneur, the use of GenAI is more than a tool for operational efficiency, it’s a bridge between technological advancement and social well-being. According to recent surveys, a sizable portion of SMEs are implementing GenAI technologies to improve customer engagement and operations (Mae Anderson, 2024). For example, according to a survey conducted by Teneo and the US Chamber of Commerce, 98% of small businesses are using AI-enabled tools, with 47% specifically using GenAI tools like chatbots and image creation (Mae Anderson, 2024). Furthermore, according to a DBS survey, 73% of SME leaders intend to spend money on GenAI solutions to increase productivity (Gemma Iso, 2025). In many developing markets, including Vietnam, small and medium-sized businesses (SMEs) are the foundation of the economy (OECD, 2023). However, in contrast to big corporations, these companies frequently do not have access to sophisticated infrastructure, a wealth of resources or comprehensive customer analytics. GenAI makes everything equal and gives SMEs the ability to provide individualized services, encourage environmentally friendly operations and even involve underserved populations, such as senior citizens who require easier-to-use interfaces or individuals with disabilities who require specialized support (Modgil et al., 2025). Although GenAI demonstrates substantial technological potential, the broader social effects of these advancements on consumers remain insufficiently understood. When using AI-augmented services, how do users feel? Do they appreciate and acknowledge these advantages? And do customers think that GenAI used by SMEs as a springboard to accomplish more general social objectives like Sustainable Development Goal (SDG 3) (Good Health and Well-Being or Positive Social Impact) of the UN?

Social impact has emerged as a critical outcome in the context of technology-enabled service innovation, particularly for SMEs leveraging GenAI. Drawing from the use of GenAI in marketing, which uses strategic, evidence-based marketing approaches to improve business (Dwivedi et al., 2023), this study views GenAI-enabled services in SMEs as mechanisms not only for business value creation but also for fostering social well-being. These AI-driven services, ranging from personalized customer to inclusive digital access (Acar, 2024), have the potential to shape behaviors, support vulnerable populations and contribute meaningfully to social and emotional outcomes in local communities (Baabdullah, 2024). However, the knowledge space regarding the antecedents of how customers perception toward social impact when SMEs used GenAI is limited and thus remains under-explored.

With service-dominant logic (SDL) as a theoretical lens on hybrid intelligence, the interaction between businesses, customers and AI can be seen as a value co-creation process, where all actors contribute and integrate resources (Simón et al., 2024). In AI-powered personalized customer service, customers provide contextual information, feedback and preferences, while AI models process this data, generate personalized recommendations and enhance customer interactions (Simón et al., 2024). A key aspect of SDL is the co-creation of value, emphasizing the collaborative development of services where both customers and AI actively participate in mutually beneficial resource integration. Following value co-creation theories, customers are not merely passive recipients of AI-driven services; rather, they are active participants in shaping and refining AI interactions (Saha et al., 2020). Through continuous engagement, such as feedback loops and preference updates, customers help improve AI accuracy and service relevance, leading to enhanced well-being, whether by reducing cognitive load, minimizing search effort or receiving highly tailored assistance. The unique nature of SDL lies in its value-in-use perspective (Nadeem et al., 2021; Ranjan and Read, 2016), which highlights the subjective benefits customers experience during interaction rather than merely consuming a predefined AI-generated solution. Value co-creation occurs within a joint sphere of human and machine intelligence (S. Li et al., 2021), where AI augments human capabilities while customers and employees contribute insights to improve AI-driven services as depicted in Figure 1. Thus, applying a value co-creation and value-in-use approach, businesses can design AI-enabled customer service systems that enhance well-being, foster personalized engagement and create adaptive, human-centered AI solutions (Amin and Di Pietro, 2025). This dynamic interaction implies that value is cocreated at a single point of interaction, reinforcing the role of AI as a facilitator of both customer satisfaction and overall well-being. With the importance of GenAI in SMEs to facilitate the value co-creation process, the perception of customers toward how GenAI used in SMEs and the customer’s belief how their voice matters are crucial to boost their willingness to participate in co-creation (Bandura, 1986; Hasan et al., 2024), which then results in greater value-in-use and positive social impact.

Reviewing the prior literature on GenAI, value co-creation, the study identified some research gaps. First, existing literature has focused primarily on adoption and implementation of GenAI from a business perspective, providing a limited insights on perception and experience of customers when using GenAI-powered products and services. In other words, numerous previous studies have explored how GenAI supports to businesses and drives their adoption (Creely and Blannin, 2025; Fu et al., 2025; Teng et al., 2025), while few research has paid enough attention to how GenAI helps businesses influence customers’ well-being and contribute to positive social impact. Second, although scholars realize that GenAI provides opportunities for SMEs to offer personalized products and services to customers (Fu et al., 2025), allow employees to focus on customer value (Teng et al., 2025) and reduce the operational cost (Creely and Blannin, 2025) then improve the positive social impact (Emerald Publishing, 2025; Zhu and Dolnicar, 2025). However, there is no evidence on how customers actually perceive and experience social benefits-especially in the context of SMEs. Third, as seen through the lens of SDL, value is not built into AI technology itself, but rather in the interactions between small businesses and their customers (Saha et al., 2020). This point of view focuses on value-in-use, which means that customers get unique value from using GenAI services in their own situations, like getting personalized help, faster service or emotional support. SDG3 calls for people to be healthy and happy, and these experiences can help people reach that goal (Emerald Publishing, 2025). However, there is not a lot of real-world evidence on how customers get this kind of value-in-use or how small businesses use GenAI to improve social well-being. As the result, there is a need to investigate how GenAI adoption, cocreating value with customers and social impact outcomes all work together.

To address these gaps, the present study investigates how customers perceive and engage with GenAI-enabled services offered by small and medium-sized enterprises (SMEs). GenAI, as a subset of AI and machine learning, refers to systems capable of autonomously generating new and contextually relevant outputs, such as text, images and solutions, based on learned patterns rather than predetermined rules (M. Kim, 2023). SMEs, defined here according to European criteria, are firms using fewer than 200 workers with an annual turnover below VND 200bn (European Commission, 2021). This study simultaneously examines key psychological factors, perceived knowledge of GenAI and perceived marketplace influence, that drive customer engagement in value co-creation behaviors. The study also introduces value-in-use and emotional well-being as critical antecedents of social well-being perception, marking the first empirical effort to link cognitive-motivational mechanisms to well-being outcomes in AI-driven SME services. By integrating SDL with SDG 3 (Good Health and Well-being), this study conceptualizes customers as active cocreators of value rather than passive recipients of technology and answers the following:

This integration bridges microlevel consumer experiences (emotional fulfillment and perceived empowerment) with macrolevel social objectives (collective well-being and inclusive service ecosystems), thus offering a multilevel framework for assessing social impact. Furthermore, by concurrently examining cognitive (perceived knowledge) and normative (marketplace influence) antecedents, the study delineates dual pathways through which customers engage in value co-creation within GenAI-enabled SME environments.

This article initially introduces a theoretical model and the development of hypotheses. The research methodology, results, conclusion, implications and limitations are subsequently presented.

Recent studies have expanded understanding of GenAI beyond functional efficiency toward its emotional and social dimensions in service contexts. Yoon et al. (2025) emphasized both the bright and dark sides of GenAI services, noting how affective and trust-related mechanisms shape user well-being and social trust. Similarly, Creely and Blannin (2025) explored human–AI co-creativity, showing that emotional design and perceived authenticity influence engagement and acceptance. In organizational and nontechnological settings, Modgil et al. (2025) and Teng et al. (2025) demonstrated that GenAI’s value creation increasingly depends on users’ perceived psychological comfort, empathy and relational trust with AI interfaces. Together, these studies highlight that emotional resonance and perceived trustworthiness are critical antecedents of customer engagement and value co-creation in GenAI-enabled service experiences, providing an essential foundation for the present research’s focus on social well-being outcomes in SME services.

Perceived knowledge refers to the beliefs consumers hold regarding their understanding of a particular issue (Wilcock et al., 2004). In this study, perceived knowledge regarding GenAI in SMEs refers to the degree to which customers believe they understand how small and medium-sized enterprises (SMEs) apply GenAI technologies in designing, producing and delivering personalized products or services that may affect their satisfaction and well-being. Perceived knowledge significantly influences risk perceptions and behavioral intentions (Liao et al., 2020). The SDL framework suggests that co-creation necessitates resource integration, with user knowledge serving as a crucial operand resource that influences and determines service outcomes (Vargo and Lusch, 2008). In addition, the technology acceptance model and social cognitive theory suggest that knowledge enhances users’ ability to engage effectively with technological services, thereby increasing their propensity to participate in co-creation (Bandura, 1986). Various studies have demonstrated a significant correlation between customer perceived knowledge and behavior (Joshi and Rahman, 2019; Liao et al., 2020; Wilcock et al., 2004). Recent studies further indicate that customers who comprehend how GenAI functions are more likely to provide feedback, personalize interactions and engage actively with AI-driven interfaces, behaviors consistent with value co-creation (Creely and Blannin, 2025; Fu et al., 2025; Teng et al., 2025). As a result, we propose the following hypothesis:

Social perception significantly influences judgment and behavior (Leary et al., 2014). In the context of sustainable consumption, perceived marketplace influence refers to the belief that an individual’s actions in the marketplace can affect the behavior of other consumers and organizations, motivating their own engagement (Leary and Vann, 2016). From the perspective of the SDL, when consumers recognize that their purchasing decisions can impact market trends and business practices, they are more inclined to engage actively in the development of services or products (Vargo and Lusch, 2004). This influence promotes behaviors including feedback provision, offering customization and codeveloping solutions, essential activities in value co-creation. Consequently, perceived marketplace influence is defined as customers’ belief that their individual purchasing decisions and engagement behaviors, particularly toward products and services powered by GenAI, can influence how small and medium-sized enterprises (SMEs) design, produce and deliver their offerings. This construct reflects customers’ perceived ability to shape marketplace outcomes by supporting technology-driven innovations that align with widely recognized principles of social responsibility, such as ethical use of AI, transparency, inclusivity and sustainability (Mohr et al., 2001). Previous studies have established a positive correlation between perceived marketplace influence and customer engagement in the purchasing process of products or services beneficial to society and the environment (Leary et al., 2014; Leary and Vann, 2016). Following the study of Batat and Hammedi (2023), customer perception of the significance of their input enhances motivation to participate in co-creation activities. This belief fosters proactive engagement, including the provision of reviews, feedback and interaction with AI tools that enhance products and services. As a result, we propose the following hypothesis:

According to the S-D Logic proposed by Vargo and Lusch (2008), value is not inherent in the product but is cocreated through its utilization. In S-D logic, the roles of producers and consumers are indistinguishable, indicating that value is perpetually cocreated, collaboratively and reciprocally, in interactions between providers and beneficiaries through the amalgamation of resources and the utilization of competencies (Vargo et al., 2008). This study demonstrates that value co-creation behavior, defined by customer engagement and interaction with GenAI systems, allows SMEs to provide personalized, contextually relevant services. The integration of human input and AI capabilities in enhances the perceived value-in-use of the cocreated service experience (M. M. Li et al., 2024). This aligns with S-D logic, which underscores that value is phenomenologically established in use and it illustrates how GenAI-enabled SMEs can flexibly modify offerings to meet specific customer requirements, resulting in emotional fulfillment, convenience and functional advantage (Grönroos and Voima, 2013). Customer participation in co-creation enhances the AI system’s refinement and alignment with their preferences, thereby augmenting value-in-use, the subjective perception of benefit derived from using the service. As a result, we propose the following hypothesis:

The concept of emotional well-being encompasses the degree to which customers encounter positive emotional experiences, like satisfaction, peace of mind, empowerment and trust (Uslu and Tosun, 2024), during and following their interactions with services or products provided by SMEs that incorporate GenAI technologies. In the context of GenAI adoption by SMEs, emotional well-being includes customers’ perceptions of being supported, understood and valued through personalized, responsive and intelligent service experiences facilitated by AI (B. J. Kim and Lee, 2024). The investigation conducted by Zhou et al. (2024) in social commerce reveals the connection between value co-creation and emotional well-being. Furthermore, Uslu and Tosun (2024) demonstrate that value co-creation significantly enhances customer satisfaction and emotional well-being. As small and medium enterprises use GenAI to improve service delivery, incorporating elements like automated recommendations, virtual assistance or tailored interactions, customers might encounter emotional advantages such as decreased stress, heightened confidence in their decision-making and an enhanced feeling of connection or control. The emotional responses play a significant role in shaping overall subjective well-being and affect the perception of service value (Khan et al., 2020). As a result, we propose the following hypothesis:

In the field of SMEs implementing GenAI, value-in-use can manifest through hyper-personalized services, rapid responses or astute recommendations that enhance customers’ daily experiences (Rezazadeh et al., 2025). When customers perceive that these AI-enhanced offerings fulfill their immediate needs while also fostering inclusivity, accessibility or environmental awareness, they are more inclined to associate these services with wider social advantages (El-Shihy et al., 2024). This corresponds with SDL, which posits that value is not inherent in the product or service but is cocreated through its utilization (Vargo and Lusch, 2008). Consequently, when SMEs use GenAI to collaboratively generate value with customers, the perceived personal advantage may extend to perceptions of social welfare, including equity, community empowerment or diminished information asymmetry (Kot and Leszczyński, 2022). As the result, we propose the following hypothesis:

Within GenAI-enabled services offered by SMEs, customers may experience emotional relief due to ease of access, responsiveness and personalized attention, fostering a sense of being understood and valued (Modgil et al., 2025). These emotional gains extend beyond the self, influencing perceptions of societal good. According to Lin Li et al. (2025), emotions from service experiences can inform broader judgments about fairness, societal benefit and collective progress. Hence, when customers feel emotionally supported by intelligent services, they may infer that these innovations contribute positively to societal well-being, promoting trust in technology’s role in inclusive service delivery, equitable access and mental health support (B. J. Kim and Lee, 2024; Quaye et al., 2024). Thus, emotional well-being can act as a precursor to heightened social well-being perception, reinforcing the belief that such technologies advance community-level outcomes. We propose the following hypothesis:

Based on the hypothesis’s development, the conceptual model has been presented in Figure 2.

This study’s empirical data was gathered from users through a questionnaire consisting of two main sections. The initial section examines essential consumer demographics, including gender, age, education level, occupation and income level. Section 2 corresponds with the research model and includes six components. To ensure the effectiveness of all instruments, each construct was measured using multiple items ( Appendix 1), all adapted from prior studies and modified to fit the context of GenAI use in SMEs. The questionnaire uses a seven-point Likert scale, from 1 (strongly disagree) to 7 (strongly agree), for the evaluation of all items. Value Co-creation Behavior items were derived from Yi and Gong (2013), emphasizing customer participation and feedback engagement. Value-in-use items are derived from the works of Vargo and Lusch (2008), highlighting both practical and affective utility. Emotional well-being was assessed by modifying items from Uslu and Tosun (2024), whereas the perception of social well-being was informed by contemporary frameworks in AI-enabled service research (Colenberg et al., 2020; Dey and Gibbon, 2017). Perceived knowledge about GenAI in SMEs and perceived marketplace influence adapted from Liao et al. (2020) and Joshi and Rahman (2019), respectively. All items underwent pilot testing with 30 respondents to verify clarity and reliability prior to extensive distribution.

Partial least squares structural equation modeling (PLS-SEM) was used to evaluate the theoretical model presented in this study. The target population consisted of Vietnamese consumers who have engaged with or made purchases from SMEs using GenAI-powered services, such as AI chatbots, AI-generated content and AI-based personalization. Vietnam was selected as the research context because it represents one of Southeast Asia’s most dynamic emerging economies, where SMEs account for over 97% of total enterprises and are rapidly adopting AI technologies to enhance service delivery and customer engagement (Thi Viet and Nguyen, 2025). The Vietnamese government’s National Digital Transformation Program (2025–2030) has further accelerated the integration of AI and automation in SME operations, making the country an appropriate and theoretically relevant context to explore how GenAI-driven services influence customer perceptions, engagement and well-being (Jamie Nguyen, 2024; Phi Nhat, 2024). A purposive sampling method was used to recruit participants who possessed knowledge of or experience with GenAI in service contexts. Data were gathered using an online survey disseminated through social media platforms, professional networks and consumer forums centered on AI and digital innovation in Vietnam. Participants were required to meet the following criteria:

• be at least 18 years of age;

• have made a purchase from a small or medium-sized enterprise using AI technology within the past 12 months; and

• be residents of Vietnam.

The sample size emerged as a significant concern due to the application of structural equation modeling to evaluate the mentioned assumptions. G-power (version 3.1.9.7) determined that the minimum sample size necessary for PLS-SEM analysis is 102 (Duc et al., 2024; Phan et al., 2025). A total of 180 responses were obtained, fulfilling the minimum sample size requirement.

Prior to participation, all respondents were informed about the purpose of the study, the voluntary nature of participation and the confidentiality of their responses. Participants provided informed consent before completing the questionnaire and were informed that they could withdraw from the survey at any time. Ethical approval for this study was obtained from the university ethics review board prior to data collection.

The demographic details of the respondents are shown in Table 1. Of the respondents, 63.33% are women and 36.67% are men. The majority of responders (69.44%) were in the 18–24 age range. Furthermore, the majority are educated; almost 72.28% of them hold a bachelor’s degree or higher. Most of them are working with more than 62.78% of respondents. A total of 77.22% of respondents claimed their monthly income is above 5 million.

Survey surveys may exhibit common method bias when the same person answers all questions (Podsakoff et al., 2003). To reduce method bias, variables are randomly placed on questionnaire pages to reduce order effects and discourage respondents from detecting causal correlations (Hulland et al., 2018). Two statistical analyses were undertaken to evaluate the severity of the common method bias. Using Harman’s single factor test, the study found that a single common factor explained 42.597% of variation, which was below the 50% threshold (Ooi and Tan, 2016; Wang et al., 2022; Wong et al., 2023). We found that the variance inflation factor (VIF) of all variables in the research model ranged from 1.000 to 3.298. The VIF score below 3.3 shows no structural model collinearity (Hew et al., 2020; Lim et al., 2021; Yuan et al., 2021).

Assessing Cronbach’s alpha (α), item loading, composite reliability (CR) and average variance extracted (AVE) values determined the reliability and validity of the research variables. Table 2 demonstrates that all constructs have strong internal reliability, with Cronbach’s α and CR exceeding the threshold of 0.7 (except PEK4 and PEM1, however, the AVE of PEK and PEM are 0.5 (Hair et al., 2011; Tan and Ooi, 2018), the study still keeps these items). For convergent validity, we set a threshold of 0.7 item loading and AVE of 0.5. Table 2 also shows that all item loadings and constructs’ AVE are above the indicated threshold values, indicating high convergent validity.

Researchers evaluate discriminant validity using the heterotrait-monotrait (HTMT) ratio (Henseler et al., 2015). Table 3 reveals that the HTMT ratios of all constructs are below the acceptable threshold of 0.90. Results suggest our research model has strong discriminant validity.

A structural model analysis is performed to assess the research model’s R², Q², f² and postulated path linkages. To measure the model’s explanatory power, the determinant coefficient R² was assessed first. R² values of 0.236 indicate the model explains 23.6% of the variance in social well-being perception, showing acceptable explanatory power. The Stone–Geisser Q² value was derived using blindfolding to assess the model’s predictive significance. Hair et al. (2014) found that the route model’s prediction accuracy is satisfactory, as the Q² value of SOW is 0.196. This study assessed impact size f² using Cohen (2013) guidelines, including 0.02, 0.15 and 0.35 indicating small, medium and large effects. Tables 4 and 5 summarize tested determinant coefficient (R²), predictive relevance (Q²) and effect size (f²). Research hypotheses are tested using SmartPLS 4.0 and PLS-SEM. To test the significance of path coefficients in the research model, we used the bootstrapping subsampling technique (5,000 subsamples). Table 6 and Figure 3 indicate that five of six hypotheses for direct effects are supported based on direct relationship test findings except H6. In particular, PEK (β = 0.490, t = 6.557, p < 0.001) and PEM (β = 0.328, t = 4.244, p < 0.001) significantly drive VAC, supporting H1 and H2. Moreover, VAC drives VAI (β = 0.749, t = 14.507, p < 0.001) and EMW (β = 0.762, t = 19.536, p < 0.001), supporting H3 and H4. Finally, VAI (β = 0.425, t = 4.356, p < 0.001) substantially correlates with SOW, H5 is supported. However, EMW (t = 1.053, p > 0.05) unexpectedly correlated with SOW. So, H6 is not supported.

The structural model results support five of the six proposed hypotheses. For H1, the positive and significant relationship between perceived knowledge about GenAI and value co-creation behavior highlights that informed customers are more likely to engage in co-creation. This builds on prior research by Fu et al. (2025) and Teng et al. (2025), which primarily focused on the business side of GenAI adoption, by shifting attention toward customer engagement and agency. It also reinforces SDL, emphasizing that knowledge empowers consumers to become value cocreators (Vargo and Lusch, 2008). Similarly, H2 supports the notion that perceived marketplace influence is positively associated with value co-creation behavior – an insight that extends beyond Hammedi et al. (2024), who focused on the voice of the consumer, by showing how normative beliefs about market impact can also drive participatory behavior. In line with previous literature emphasizing the experiential nature of value (Grönroos and Voima, 2013), H3 confirms that value co-creation is positively associated with value-in-use. This result strengthens the understanding that collaborative interactions between customers and GenAI-enabled SMEs lead to enhanced personalized service outcomes. Furthermore, H4 highlights the emotional benefits of co-creation, aligning with affective events theory (Weiss and Cropanzano, 1996) and extending the findings of Kim (2023), who observed emotional satisfaction in service innovation contexts, but without the GenAI lens. The relationship between value-in-use and social well-being offers a novel contribution by linking microlevel service outcomes with macrolevel societal perceptions, a connection that remains underexplored in GenAI literature (Creely and Blannin, 2025; Emerald Publishing, 2025). Finally, the results for H6 indicate that emotional well-being is not positively associated with perceptions of social well-being, providing an interesting insight to the broader validation of SDG 3 (Good Health and Well-being) in AI-powered service contexts. While emotional well-being contributes to individual satisfaction and psychological fulfillment, its effect may not extend to how customers evaluate collective or societal well-being. This outcome could stem from measurement boundaries that capture primarily individual affective states rather than socially oriented outcomes, or from cultural tendencies that distinguish personal emotions from community-based judgments. It may also reflect a theoretical misalignment, as affective responses often occur at an individual level, whereas social well-being reflects evaluative, relational and ethical considerations. These findings suggest that SMEs should make their societal and community contributions more visible and explicitly connect personalized GenAI services with inclusive, transparent and socially responsible practices.

This study makes several important contributions to the literature on GenAI, value co-creation and social impact within the context of SMEs. First, this study advances the social impact literature by providing empirical evidence on how customers experience and derive value from GenAI-enabled services. It establishes that social impact in AI-mediated SME contexts is not solely a technological outcome, but a human-centered process shaped by customer perceptions, engagement and co-creation behaviors (Baabdullah, 2024; Dwivedi et al., 2023). The findings show that GenAI can serve as a social enabler when customers perceive transparency, control and personalization, reinforcing the idea that responsible AI deployment in SMEs can enhance community trust and shared well-being. Second, by integrating perceived knowledge and perceived marketplace influence as antecedents of value co-creation, this study enriches SDL theory (Vargo and Lusch, 2004, 2008) with a dual-path cognitive–normative model that explains why customers engage with GenAI-driven services. The results reveal that customers’ sense of agency and understanding of AI functions jointly motivate participation in value creation, underscoring that value-in-use emerges through meaningful human and AI collaboration rather than technological automation alone (Simón et al., 2024). Third, the nonsignificant relationship between emotional well-being and social well-being offers a nuanced theoretical contribution. It suggests that while GenAI interactions may enhance individual emotional satisfaction (e.g. personalization and convenience), they do not automatically translate into perceived social connectedness or collective benefit. This finding highlights a “human disconnect” in GenAI-mediated service encounters, where emotional gratification remains individual rather than communal, signaling a need for SMEs to design AI experiences that visibly foster inclusion, empathy and societal contribution (Lin Li et al., 2025; Modgil et al., 2025). Finally, drawing on the Yi and Gong (2013) co-creation framework, this study identifies that behaviors such as providing feedback, adapting preferences and cooperating with AI interfaces contribute more strongly to social well-being perceptions than mere information sharing. These behaviors embody the “human touch” in GenAI contexts, demonstrating that meaningful social value arises not from the algorithm itself, but from customers’ willingness to humanize their interactions with AI-enabled services (Saha et al., 2020; Simón et al., 2024).

From a practical standpoint, the findings offer concrete guidance for SME service owners seeking to leverage GenAI for both business growth and positive social outcomes.

First, SMEs should proactively inform customers about how GenAI is used in their services, clarifying what the AI does, how it personalizes experiences and how data are protected. As demonstrated, higher perceived knowledge of GenAI increases customers’ willingness to cocreate. Simple initiatives such as short explainer videos, transparent chatbot disclosures or interactive tutorials can enhance trust and engagement.

Second, among the four co-creation behaviors identified by Yi and Gong (2013), information sharing, responsible behavior, personal interaction and feedback provision, this study indicates that feedback provision and personal interaction most strongly contribute to customers perceived social well-being. SME owners should therefore design GenAI touchpoints that invite two-way dialogue, allow for feedback customization and promote human–AI collaboration rather than automation alone. For example, allowing customers to codesign product features or provide real-time feedback on AI recommendations can heighten feelings of inclusion and collective benefit.

Third, SMEs can strengthen customers’ sense of agency by highlighting how their participation drives ethical, inclusive or sustainable outcomes. This can be achieved through participatory design platforms, social media acknowledgments of customer input or reward programs linked to socially responsible choices. When customers perceive that their actions shape business practices, they are more likely to sustain co-creation behaviors that foster shared well-being.

Finally, beyond efficiency, GenAI should be designed to serve human needs, emotional satisfaction, empathy and accessibility. Personalized recommendations, empathetic AI tone and multilingual or inclusive interfaces can promote emotional well-being, which in turn supports customers’ perceptions of social value and fairness. To enhance the practical applicability of the findings, this study proposes a SME GenAI social impact implementation checklist ( Appendix 2). The checklist translates the empirical findings into actionable managerial guidelines that help SME owners evaluate whether their GenAI-enabled services effectively support customer co-creation, emotional well-being and social impact outcomes. The framework is grounded in the study’s core constructs, including perceived knowledge, perceived marketplace influence, value co-creation behavior, value-in-use, emotional well-being and social well-being perception. The checklist can be used periodically as an internal monitoring and continuous improvement tool. In addition, policymakers and SME support organizations may adapt the framework into AI-readiness assessment programs, training workshops or digital transformation guidelines to promote responsible and socially beneficial GenAI adoption among SMEs.

Government agencies, chambers of commerce and industry associations can use these insights to create SME-friendly AI literacy programs and co-creation toolkits. Establishing clear ethical standards and community feedback channels will ensure that GenAI adoption aligns with national social and sustainability goals.

Despite its contributions, this study is not without limitations. First, the use of purposive sampling within the Vietnamese context, while appropriate for capturing insights from customers familiar with GenAI-enabled SME services, may constrain the generalizability of findings across different cultural or economic environments. Future research should expand the sampling scope to multiple regions or countries to validate the model’s applicability and strengthen external validity. Second, the cross-sectional design limits the ability to infer causal relationships over time; hence, longitudinal or experimental approaches are recommended to examine how customer perceptions and value co-creation behaviors evolve with sustained engagement in GenAI-enabled services. Third, although demographic characteristics such as age, gender and education were reported, they were not included as control variables in the structural model to preserve theoretical parsimony. Future studies could incorporate these controls to test potential moderating effects and verify the robustness of the model across customer subgroups. Fourth, while the study integrated cognitive (perceived knowledge) and normative (marketplace influence) factors, it did not capture affective or relational dimensions such as trust in AI, perceived empathy or AI–human emotional alignment. Including these constructs could offer a more comprehensive understanding of how users emotionally and socially interact with GenAI-driven services. Fifth, the model focused primarily on customer-level perceptions and experiences, potentially overlooking the organizational and technological readiness of SMEs. Future research could adopt a dyadic perspective that considers firm-side capabilities and constraints to better reflect the multidimensional dynamics of value co-creation within AI-enhanced SME ecosystems. Sixth, the construct of perceived knowledge captures customers’ subjective understanding of GenAI technologies rather than their actual expertise. While this perception is central to behavioral intention, discrepancies between perceived and actual knowledge could lead to ineffective or even counterproductive co-creation behaviors, particularly in high-stakes service contexts such as finance or healthcare. Future studies should therefore differentiate between perceived and objective knowledge to examine how knowledge accuracy moderates the quality and outcomes of co-creation across different types of GenAI-enabled services. Finally, although perceived marketplace influence was hypothesized to have a positive effect on value co-creation, its impact may not always be uniformly beneficial. Customer perceptions of influence may, under certain conditions, result in adverse outcomes, for example, when GenAI-generated outputs produce unanticipated errors that disadvantage either the customer or the SME. This suggests that the valence of marketplace influence is likely context-dependent, shaped by perceived fairness, transparency and alignment between customer and firm interests. Future research could adopt a more nuanced approach to explore the bidirectional nature of marketplace influence and its ethical implications in AI-mediated service interactions.

During the preparation of this work, the author(s) used ChatGPT-4 for editorial purposes, including grammar correction, language enhancement and improving the overall clarity of the manuscript. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.

Managerial interpretation guide:SME managers can use this checklist as a practical self-assessment tool to evaluate the social readiness of their GenAI-enabled services. Each item should be scored on a five-point scale, where 1 indicates “not implemented” and 5 indicates “fully implemented and consistently practiced.” A total score between 40 and 50 suggests a highly developed human-centered GenAI strategy with strong social impact orientation. Scores between 25 and 39 indicate moderate readiness, where AI systems support customer engagement but still require improvements in transparency, empathy or participatory design. Scores below 25 suggest that GenAI adoption remains primarily technology-driven and may not yet generate meaningful social well-being outcomes for customers. The checklist can also function as a continuous improvement framework. SME managers may conduct periodic evaluations (e.g. quarterly or biannually) to monitor whether their AI-enabled services increasingly promote customer trust, co-creation, inclusiveness and perceived social value. In addition, policymakers, chambers of commerce and SME support agencies may adapt this framework into training programs or AI-readiness toolkits to encourage responsible and socially sustainable GenAI adoption among SMEs.