Social impact creation with GenAI in small- and medium enterprises in healthcare Open Access

Generative artificial intelligence (GenAI) – which is defined as a type of artificial intelligence (AI) that can create new content, such as text and images – is omnipresent with ChatGPT, Google Gemini and Microsoft Copilot as popular GenAI tools (Nield, 2025). Across sectors and industries, individuals and organizations increasingly rely upon GenAI tools for the delivery of services (Sigala et al., 2024), with professional service providers reporting adoption rates of around 80% across functions (Statista, 2025). The growing diffusion of GenAI in services has sparked optimism regarding its potential to enhance organizational well-being through improved performance and efficiency of the service delivery processes (e.g. Grigsby et al., 2025; Huang and Rust, 2024), as well as to contribute to the well-being of stakeholders and broader service ecosystems (e.g. Alkire et al., 2024; Sidaoui et al., 2024).

At the same time, integrating GenAI into service delivery systems is associated with substantial technical, financial, regulatory and ethical issues (e.g. Belanche et al., 2024; Gupta and Rathore, 2024; Kronblad et al., 2024; Sigala et al., 2024). In other words, the GenAI literature presents a contradiction: on the one hand, GenAI is portrayed as a driver of efficiency and well-being in service organizations; on the other hand, its implementation has been framed as a source of risk and complexity. These opposing perspectives point to an unresolved tension that requires reconciliation among service organizations considering the implementation of GenAI, particularly healthcare organizations and small and medium-sized enterprises (SMEs) delivering services under resource constraints (Dolmans et al., 2014; Van Zyl et al., 2021).

When service organizations operate in low-resource settings – i.e. contexts with systemic, multidimensional resource limitations (Van Zyl et al., 2021), we argue that the stakes of integrating GenAI are amplified. Such organizations stand to benefit the most from the efficiency and well-being gains that GenAI promises, yet they simultaneously have fewer resources to manage the associated risks (Dolmans et al., 2014; Van Zyl et al., 2021). Meanwhile, extant research suggests that innovations – such as the integration of GenAI in organizations – are both enabled and constrained by organizational resource limitations (Dolmans et al., 2014; Kraaijenbrink et al., 2010). Given these inconsistencies in the literature, it remains unclear how resource-constrained service organizations – i.e. service organizations facing resource limitations along multiple dimensions (e.g. financial, technological, capability, etcetera) – can effectively deal with GenAI.

To resolve the inconsistencies linked to integrating GenAI in resource-constrained service organizations, the present research investigates how integrating GenAI in these service organizations affects their well-being and that of their stakeholders – which is how social impact has been defined (Ebrahim and Rangan, 2014; Lindgreen et al., 2021; OECD, 2015). To gain insight into social impact creation with GenAI tools for service organizations with resource limitations, we focus on the healthcare sector as a prototypical low-resource setting and SMEs as resource-constrained organizations within this setting (Van Zyl et al., 2021). Indeed, healthcare SMEs aim to generate social impact (Akinola and Obokoh, 2024) but can – due to their sector and size – face specific challenges when implementing new technologies (Nguyen, 2009) like GenAI (Carayannis et al., 2024).

When it relates to integrating GenAI in the delivery of healthcare services, prior scholarly work has – as shown in Table 1 – focused on the current state of (Gen)AI research and its benefits and risks from an individual, organizational and sectoral/systemic perspective. Yet, none of these studies have specifically examined GenAI in the specific context of healthcare SMEs (rather than other types of organizations). Healthcare SMEs, which aim to enhance individual and societal well-being, are ubiquitous in the healthcare sector (Akinola and Obokoh, 2024). In fact, these organizations have the ambition to create social impact, that is, fostering not only their own well-being but also that of their stakeholders (Alkire et al., 2025; Parkinson and Naidu, 2024). Meanwhile, healthcare SMEs operate in low-resource settings, thereby turning them into resource-constrained service organizations (Nguyen, 2009; Van Zyl et al., 2021). The most common resource constraints for organizations are – according to Dolmans et al. (2014) – financial, capacity and capability constraints. These resource constraints are – as shown in Table 2 – also relevant for healthcare SMEs. A key question, however, revolves around how these resource constraints affect the integration of GenAI in healthcare SMEs.

Based on the resource-based view (Barney, 1991), access to valuable, rare, inimitable and non-substitutable resources – as well as complementary skills and capabilities – is necessary for implementing innovations like GenAI-enabled services (Kraaijenbrink et al., 2010) and cocreating value with them in service ecosystems (Crumbly et al., 2024; Vargo and Lusch, 2016). Among all resources, financial resources are often seen as the most critical ones as it allows organizations to invest in technology and human beings that boost the development of capabilities (Thokala et al., 2025; Cecere et al., 2020; Van Zyl et al., 2021) and hence contribute to the innovative capacity of organizations (Woschke et al., 2017). In other words, organizations can – as also suggested by service-dominant logic (Vargo and Lusch, 2016) – leverage these resources to develop skills and capabilities and apply them to integrate innovations. However, when resources are constrained, innovation such as the integration of GenAI-enabled services is – consistent with this evidence – less likely to occur.

Paradoxically, while the resource-based view predicts that resource constraints reduce the likelihood of innovation (Kraaijenbrink et al., 2010), resource constraints are – according to Dolmans et al. (2014) – not necessarily insurmountable and may even stimulate innovative solutions. Some researchers and practitioners contend that certain innovations – such as GenAI-enabled services – can also help organizations to overcome resource limitations by augmenting human capabilities, improving efficiency, and enabling new service delivery models and thereby potentially foster their well-being and that of their stakeholders (De Blick et al., 2024; Eurofound and Cedefop, 2025; Lee et al., 1999; Wirtz and Stock-Homburg, 2025), provided they are designed and deployed responsibly (see also Table 3 for an overview of calls for responsible GenAI integration in service organizations).

Considering these paradoxical perspectives on the role of resource constraints, we conclude that they may both inhibit and possibly benefit innovations like the integration of GenAI in service delivery processes. Yet, responsible integration of GenAI-enabled services in resource-constrained organizations requires a comprehensive understanding of its well-being implications. In other words, the well-being opportunities and risks associated with GenAI integration in resource-constrained environments warrant careful examination – particularly when these organizations aim to create social impact, as is the case for healthcare-related SMEs.

To better understand the social impact that GenAI can create for healthcare SMEs as resource-constrained service organizations, we therefore ask the following research question: how can resource-constrained service organizations, particularly healthcare SMEs, effectively integrate GenAI to maximize social impact while managing the risks inherent to its adoption? To investigate how GenAI shifts, amplifies or complicates this paradox, we conduct a systematic review of academic literature alongside business and government publications to examine the well-being opportunities and risks associated with integrating (Gen)AI into healthcare service delivery across multiple ecosystem actors. In line with the guidelines for responsible GenAI integration in service ecosystems (Scott and Mende, 2022; Alkire et al., 2024), we subsequently collaborate with healthcare SME experts to formulate the GenAI social impact framework, a carefully designed roadmap for practitioners to rigorously assess the social impact of integrating GenAI in resource-constrained organizations and maximize their potential to create meaningful social impact.

By doing so, we do not only create – as acknowledged by Aiello et al. (2021) and Alkire et al. (2025) – social impact through our collaboration with SME experts in the healthcare sector (cf. low-resource setting) but also contribute to a better understanding of the social impact of GenAI integration in healthcare SMEs (cf. resource-constrained organizations). Social impact research typically assumes deliberate resource investment as its starting point (Ebrahim and Rangan, 2014), whereas resource-constrained organizations pursue well-being goals under conditions of scarcity (Nguyen, 2009; Van Zyl et al., 2021). GenAI both sharpens and potentially reshapes this dynamic. This study, therefore, clarifies how healthcare SMEs can generate meaningful social impact through GenAI adoption under such constraints, thereby advancing a more precise understanding of social impact in settings where scarcity is the rule rather than the exception.

To gain insights into the creation of social impact through GenAI in resource-constrained organizations operating in low-resource settings (here, healthcare SMEs), we conducted a two-way systematic literature review including results up to December 2025 (see Figure 1). At the beginning of this process, the five researchers divided themselves into two teams of two and the fifth researcher acted as a liaison between the two teams.

The first team focused on peer-reviewed academic papers and systematically searched the Web of Science (WOS) database for this purpose. The literature search excluded proceeding paper, editorial material, meeting abstracts or news items and was limited to WOS categories in management or business. The results are based on keywords such as “generative AI,” “GenAI,” “ChatGPT,” “Artificial Intelligence” in connection with “healthcare” or “service” or “SME,” “small compan*,” “midsized compan*.” Due to the rapid and fundamental development of the topic GenAI over the past few years, only recent articles from 2022 onwards have been considered. After reviewing the abstracts of 316 papers, 41 articles were identified to be relevant for this research (see Figure 1 for more details) ^[1].

At the same time, the second team engaged in selecting prominent business and governmental publications. A systematic search was conducted using the terms like “GenAI in healthcare report” or “GenAI in healthcare” in Google. The review focused on reports selected based on the credibility of their institutional sources (e.g. government agencies, leading consulting firms and reputable nonprofit organizations). Thus, 20 publications have been selected (see Figure 1 for more details) ^[2].

As part of a multi-stage coding process, both sets of publications were initially coded separately by the teams. After identifying several categories of well-being opportunities and risks linked to the use of GenAI within the delivery of healthcare services, both teams decided – in agreement with the fifth researcher – to group these categories in more abstract categories. To integrate the insights from both teams, the fifth researcher focused on the similarities and differences between the two sets of well-being opportunities and risks, discussing the merging of both coding trees. In the third step, the merged coding tree was validated and redefined again by teams of two in mixed combinations, thereby realizing data and researcher triangulation.

In a final step, the research team purposively sampled practitioners with different roles, positions and expertise in relation to healthcare SMEs (maximal variation sampling), namely, (1) a caregiver delivering ambulant healthcare services as part of an SME, (2) a manager of a residential healthcare SME and (3) a consultant offering services to SMEs in healthcare ^[3]. After presenting the insights from our two-way systematic literature review to these healthcare SME experts with different perspectives, we engaged in in-depth conversations with them. Specifically, we discussed the well-being opportunities and risks associated with GenAI, the conditions under which these arise, and how healthcare SMEs can address them in pursuit of enhanced social impact. By collaborating with these healthcare SME experts – which is another form of data and researcher triangulation that enhances the credibility of our research – we were able to develop a comprehensive framework to rigorously assess the social impact of integrating GenAI in resource-constrained service organizations.

When striving for responsible GenAI integration, it is important to assess both well-being opportunities and risks that may arise from integrating GenAI in the delivery of healthcare services across different service ecosystem actors.

When considering the well-being opportunities, we distinguish – as shown in Table 4 – between patients, professionals and organizations. For each of these service ecosystem actors, well-being opportunities relate to generating enhanced healthcare performance (e.g. improved diagnostics and outcomes for patients, less burnout for professionals, and enhanced treatment quality for the organization), improved accessibility and efficiency (e.g. reduced waiting times for patients, improved teamwork among professionals, and better care planning for the organization), and their contribution to optimized support systems (e.g. virtual assistance along patient journey, improved clinical decision-making for professionals, and enhanced knowledge management in organizations). When multiple service ecosystem actors experience these well-being opportunities, GenAI bears the potential to deliver well-being at the system level and hence create social impact (e.g. improved national health outcomes, reallocation of freed-up resources, development of health intelligence).

As shown in Table 5, however, GenAI also brings well-being risks for patients, professionals and organizations. Each of these ecosystem actors was found to encounter ethical and legal issues (e.g. privacy concerns among patients, liability concerns among professionals, and accountability concerns for organizations) and stress and reluctance (e.g. resistance to AI-enabled care among patients, discomfort about AI-enabled diagnoses among professionals, and risk for AI-disfavouring culture within organizations). Next, (Gen)AI was linked to risks for different actors by altering healthcare jobs (e.g. dehumanized care experiences for patients, job insecurity for professionals, and demand for new skills within organizations) and taking over decision-making (e.g. fear of overreliance on AI among patients, restrained autonomy for professionals, and a lack of critical approaches within organizations). When these risks are scaled up from individual service ecosystem actors to service ecosystems as a whole, GenAI may bring well-being risks at the system level by compromising resource efficiency and institutional trust and hence diminish the potential to create social impact.

Building upon the well-being opportunities and risks that shape social impact creation with GenAI in healthcare, our collaboration with healthcare SME experts led to developing a GenAI social impact framework that includes – in light of the resource constraints – three stages:

1. identifying social impact priority areas;

2. exploring the GenAI social impact potential; and

3. implementing GenAI for social impact creation (see Figure 2).

(i.e. areas where the generation of well-being is at risk), healthcare SMEs can evaluate what service ecosystem actors are threatened in terms of their well-being (e.g. patients or healthcare professionals), what are the most important well-being risks they face (e.g. getting suboptimal care or experiencing high workload), and what resource constraints shape these well-being risks (financial, capacity, capability – see Table 2). The healthcare SME experts, for instance, refer to patients (cf. service ecosystem actor) getting suboptimal care (cf. well-being risk) due to a lack of trained professionals (cf. capability constraints), which is thus a social impact priority area. The same goes for a lack of documentation systems (cf. capacity constraints) that generates a high workload (cf. well-being risk) for healthcare professionals (cf. service ecosystem actors), as exemplified by this quote from a healthcare SME expert: “the stress for caregivers, because the documentation workload keeps growing. The most important thing would be a support tool for documentation” (manager).

(i.e. the well-being opportunities that GenAI can bring), healthcare SMEs can explore what well-being opportunities match the well-being risks that service ecosystem actors face while taking their resource constraints into consideration. With the help of Table 4, healthcare SME experts explored in what areas GenAI applications can alleviate resource constraints to creating social impact. For patients getting suboptimal care due to a lack of trained professionals, there are well-being opportunities at the intersection of professionals and enhanced healthcare performance in Table 4 (cf. GenAI applications that provide insight into conditions for improving care – “If I haven’t done [some treatments] for five years, it would be practical to have the process explained to me again, so I can implement it better.” – caregiver). The lack of a documentation system that creates stress for professionals, in turn, lead to well-being opportunities at the intersection of organization and enhanced healthcare performance in Table 4 (cf. GenAI applications that contribute to better operational performance), as explained by one of the healthcare SME experts:

The SME experts call for developing GenAI maturity (i.e. the organizational readiness and capability to integrate GenAI applications without inducing well-being risks). Here, it is important to bear in mind that many healthcare SMEs currently lack experience with GenAI technologies. As one expert mentioned, (“artificial intelligence hasn’t really been implemented in their settings yet [here, hospitals] – or at least not in any way that’s visible to users outside of the IT department” – consultant) and outsourcing its integration to third-party suppliers may not solve the issue:

With the help of Table 5, healthcare SMEs can assess what well-being risks the providers of GenAI applications foresee (e.g. privacy concerns among patients) and which ones they do not anticipate (e.g. concerns about professional liability). This assessment enables healthcare SMEs to reflect upon action to ensure that GenAI applications enhance the well-being of all actors in their service ecosystem, as explained by one of the healthcare SME experts: “some patients said, “No, I want to be treated by someone who can touch me, who can feel what’s going on, who can explain it to me.” And others said, “Oh no, actually I love this [GenAI-enabled service]. If I’m treated based on data, then I know I’m getting the best care.” (consultant). Being involved in the development of GenAI applications might enable healthcare SMEs – as resource-constrained organizations – to establish GenAI maturity (e.g. “If something is being developed, I’d hope that they would take small real-life providers like us and test those things with us. So, the prototypes would be tested in that environment” – manager). Taken together, healthcare SMEs can – as resource-constrained organizations operating in low-resource settings – only realize the social impact potential of GenAI by involving its providers and other service ecosystem actors that are affected by its integration to establish GenAI maturity.

This research advances theorizing about social impact creation by investigating what role GenAI can play. Theoretically, social impact creation has long been framed as an outcome of deliberate investment (Crumbly et al., 2024; Ebrahim and Rangan, 2014; Parkinson and Naidu, 2024), yet resource-constrained organizations – like healthcare SMEs – challenge this assumption by pursuing well-being goals under conditions of scarcity (Nguyen, 2009; Van Zyl et al., 2021). Building upon the resource-based view, this research demonstrates that GenAI offers opportunities to enhance the well-being of healthcare SMEs and their stakeholders. In these contexts, GenAI can act as a potential resource substitute and capability amplifier, as it enables resource-constrained organizations – like healthcare SMEs – to enhance their social impact in multiple ways (see Table 4). Yet, this research also shows that the integration of GenAI in resource-constrained service ecosystems comes with various risks for social impact creation (see Table 5). Through our GenAI social impact framework – which emerged by collaborating with healthcare SME experts (see Figure 2) – we reveal that the social impact of GenAI is contingent upon the capability to identify social impact priority areas, explore the social impact potential of GenAI, and implement the identified potential for social impact creation. In doing so, the framework reconciles the tension between GenAI as a driver of well-being (e.g. Alkire et al., 2024) and as a source of risk (e.g. Sigala et al., 2024). Moreover, it addresses the paradox that resource constraints may both inhibit and potentially motivate innovation (e.g. Dolmans et al., 2014; Kraaijenbrink et al., 2010). These insights reflect key concerns around the development of complementary capabilities in the resources-based view and service-dominant logic but advance it by conceptualizing social impact potential and GenAI maturity as interrelated domains that explain how new technologies transform well-being across service ecosystem actors.

For practitioners, the GenAI social impact framework (see Figure 2) and the corresponding guidelines (see Table 6) offer a structured tool to evaluate where and how GenAI can enable healthcare SMEs to create social impact. In the IDENTIFY stage, GenAI enables resource-constrained organizations to map social impact priority areas by determining which service ecosystem actors are at risk (Q1), the most pressing well-being risks they face (Q2), and the resource constraints shaping these risks (Q3). Building on this, the EXPLORE stage helps SMEs identify where GenAI can best augment service delivery, by matching GenAI opportunities to the identified well-being risks (Q4) and resource constraints (Q5). Finally, the IMPLEMENT stage assists in assessing GenAI-connected risks that providers may (not) foresee (Q6) and in developing concrete actions to enhance the well-being of all ecosystem actors (Q7), thereby establishing GenAI maturity. As the resource-based view suggests, SMEs must selectively deploy their limited resources to build capabilities that align with high-impact areas to generate disproportionate returns on investment. Ultimately, the framework empowers healthcare SMEs to make more strategic and responsible decisions regarding GenAI adoption, despite limited means.

Our research integrates existing knowledge from academic literature, practitioner and policy publications and SME expert insights. While this triangulation strengthens the framework’s relevance, three limitations merit attention. First, as our research focuses on SMEs in the healthcare sector, it is advisable to compare these findings with other types of resource-constrained organizations from other sectors or industries to identify specific differences. Second, although the framework introduces GenAI maturity as an important objective, specific action to achieve it remain underdeveloped. Future work could build instruments to measure and benchmark GenAI maturity across resource-constrained contexts and investigate how it is established. Third, our framework is based on conceptual and qualitative data, making it exploratory in nature. Future research should empirically test its applicability through case studies, ethnographic fieldwork or action research in resource-constrained organizations integrating GenAI applications for social impact creation.

The authors would like to thank the founders and organizers of Let’s Talk About Service (LTAS) 2024 for facilitating this research project. Authors are listed in reversed alphabetical order, as they contributed equally to this article. During the preparation of this manuscript, the authors used GenAI tools to check the spelling and writing of some of the included sentences. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.