Artificial intelligence, digital transformation and green human resource management in Sub-Saharan Africa: a systematic literature review

Sub-Saharan Africa (SSA) stands at an inflection point in its organisational, environmental and digital development trajectories. Home to the world’s youngest and most rapidly urbanising population, the continent faces disproportionate exposure to climate-related economic shocks, biodiversity loss, water scarcity and land degradation, risks that are projected to intensify significantly under prevailing emissions trajectories (IPCC, 2022). At the same time, SSA hosts some of the world’s fastest-growing economies, expanding formal sectors and accelerating digital adoption curves, driven principally by mobile internet penetration that has leapfrogged fixed-line infrastructure in many national contexts (GSMA, 2023). This convergence of environmental vulnerability, demographic dynamism and digital opportunity creates both an urgent imperative and a historically distinctive context for the pursuit of sustainable organisational management practices.

Green human resource management (GHRM) within this landscape has emerged as a strategically significant framework for embedding environmental sustainability within the core functions of organisations. Broadly defined as the application of HRM policies and practices that promote environmentally responsible behaviour at individual, organisational and societal levels, GHRM operationalises sustainability through green recruitment, green training and development, green performance management, green compensation and rewards and green employee involvement (Ahmad, 2015). Evidence consistently demonstrates that well-designed GHRM systems generate meaningful reductions in organisational environmental footprints, cultivate pro-environmental workplace cultures and improve sustainability-related competitive positioning outcomes of particular relevance in SSA, where regulatory and market pressures for corporate environmental responsibility are intensifying (Kramar, 2022).

Concurrently, artificial intelligence (AI) and associated digital transformation technologies are reshaping the architecture of HRM globally. AI-powered platforms, big data analytics, machine learning-driven performance monitoring systems, natural language processing (NLP)-based recruitment tools and mobile learning management systems (LMS) collectively offer unprecedented capabilities for enhancing the precision, reach, scalability and intelligence of HR functions (Charlwood and Guenole, 2022; Tambe, Cappelli and Yakubovich, 2019). Moreover, AI presents a transformative opportunity for GHRM, enabling real-time green performance monitoring, personalised environmental training at scale and AI-mediated green recruitment that identifies candidates with required skills and knowledge, all within a digital ecosystem that can dramatically reduce the environmental footprint of HR operations. However, the deployment of AI in GHRM cannot be understood or evaluated in isolation from the structural conditions of the organisational contexts in which it occurs. SSA organisations operate in environments characterised by significant digital infrastructure deficits, including unreliable broadband connectivity, intermittent electricity supply, limited cloud computing access and high technology licensing costs that substantially constrain the viability of AI-enabled HR systems (Asongu and Nwachukwu, 2018). Moreover, they are further constrained by digital skills gaps among HR professionals, nascent data protection and AI governance frameworks and the persistent risk that AI tools predominantly developed and trained in Global North institutional contexts embed cultural biases, linguistic assumptions and socioeconomic proxies that systematically disadvantage African populations (Birhane, 2021).

Despite the manifest significance of this intersection, the scholarly literature on AI-enabled GHRM in SSA remains strikingly underdeveloped. GHRM scholarship is disproportionately anchored in Asian, particularly Southeast Asian and European, contexts, with Africa accounting for a small fraction of empirical studies (Yong et al, 2020). Digital HRM scholarship, while growing rapidly, rarely engages with the distinctive structural conditions of African organisational environments (Strohmeier, 2020). The integration of AI, digital transformation and GHRM perspectives within a framework specifically sensitive to SSA institutional realities thus constitutes both a significant gap and a high-priority scholarly contribution. This systematic literature review addresses this gap by synthesising peer-reviewed scholarship published between 2018 and 2025 at the intersection of AI, digital transformation and GHRM in SSA. It is guided by two research questions (RQs):

The review makes three principal contributions to theory and practice. Firstly, it provides the first systematic synthesis of AI-enabled GHRM scholarship specifically focused on SSA, establishing an empirically grounded knowledge base for the field. Secondly, it proposes a contextually extended AMO framework that specifies the threshold boundary conditions under which AI-AMO-GHRM relationships obtain in developing-world institutional environments, offering a theoretically novel and empirically testable model. Thirdly, it integrates decolonial AI theory into GHRM discourse, foregrounding algorithmic justice, participatory design and epistemic equity as prerequisites for ethically legitimate sustainable HRM in the region and advances a future research agenda oriented toward longitudinal, comparative and participatory scholarly engagement with SSA’s distinctive organisational realities.

GHRM encompasses the systematic application of HRM policies and practices that promote environmentally responsible behaviour within and beyond organisations (Renwick et al., 2013). Operationally, it addresses five principal HR sub-domains: green recruitment and selection, which prioritises candidates with pro-environmental values and competencies; green training and development, which builds environmental knowledge, skills and attitudes; green performance management, which embeds environmental KPIs within appraisal and feedback systems; green compensation and rewards, which incentivises pro-environmental behaviours; and green employee involvement, which creates participatory channels for environmental initiative and governance (Ahmad, 2015; Tang et al., 2018).

The dominant theoretical lens in GHRM scholarship is the ability–motivation–opportunity (AMO) model (Appelbaum et al., 2000), adapted for environmental sustainability contexts by Renwick et al. (2013). The AMO model posits that effective GHRM enhances employees’ environmental abilities (green knowledge and skills), motivations (pro-environmental attitudes and incentives) and opportunities for green participation and empowerment. This multi-dimensional model has generated a productive body of empirical research demonstrating that GHRM practice bundles, particularly combinations of green training, green performance appraisal and green employee involvement, are significantly and positively associated with organisational environmental performance (Tang et al., 2018; Ren, Tang and Jackson, 2021).

Complementary theoretical perspectives enrich the AMO foundation. The natural resource-based view (NRBV), proposed by Hart (1995) as an extension of the resource-based view, positions green HR capabilities encompassing pollution prevention, product stewardship and sustainable development competencies as distinctive sources of environmental competitive advantage that are difficult to imitate and socially complex. Institutional theory, as elaborated by DiMaggio and Powell (1983), highlights how regulatory, normative and cognitive institutional pressures shape the organisational adoption of GHRM practices, explaining significant cross-national variation in GHRM prevalence that cannot be accounted for by firm-level factors alone. Social exchange theory provides a micro-foundational account of how GHRM creates reciprocal green commitment between employer and employee, with GHRM signals of organisational environmental values eliciting pro-environmental organisational citizenship behaviour (OCB-E) from employees in a social exchange dynamic (Latan et al., 2018).

Cross-contextual evidence from developing-world studies corroborates the core AMO-GHRM relationships documented in Global North scholarship. Zaid, Jaaron and Bon (2018) demonstrated in Palestinian manufacturing firms that GHRM practice bundles significantly predicted environmental and operational performance improvements, while Yusoff et al. (2020) showed in Malaysian hospitality settings that green training moderated the relationship between environmental awareness and green employee behaviour, findings with evident relevance to SSA’s manufacturing and services sectors. These cross-contextual validations suggest that the core AMO logic of GHRM holds broadly, while the mechanisms, magnitudes and enabling conditions of GHRM effects vary significantly across institutional environments.

Digital transformation is defined as the organisational integration of digital technologies across all functions to fundamentally alter how organisations operate and create value (Vial, 2019) and has progressed through several waves within HRM: from early e-HRM systems that digitalised administrative HR processes, through sophisticated data analytics platforms that enabled evidence-based workforce planning, to the current frontier of AI-driven systems that enable predictive modelling, algorithmic decision-making and real-time workforce intelligence (Strohmeier, 2020).

The application of AI to GHRM has been theorised most systematically by Ren et al. (2021), who proposed an extended AMO framework in which AI technologies function as systemic amplifiers of each AMO dimension. In the ability dimension, AI-powered LMS deliver personalised, adaptive environmental education programmes that assess comprehension in real time and recommend targeted resources based on identified competency gaps. In the motivation dimension, real-time environmental performance dashboards and AI-driven incentive optimisation systems strengthen the contingency between green behaviour and reward, enhancing motivational salience. In the opportunity dimension, digital collaboration platforms and AI-facilitated idea management systems broaden channels for employee green participation, reducing the structural barriers to environmental initiative that characterise traditional hierarchical HR systems. This extended AMO conceptualisation implies that the impact of AI on GHRM is potentially multiplicative rather than merely additive, as each AMO component is simultaneously strengthened and their interactions are facilitated by data-driven feedback loops.

The critical theoretical limitation of the extended AMO model is its implicit assumption of adequate digital infrastructure and organisational AI capability. Where these preconditions are absent as they frequently are in SSA, AI tools may fail to enhance ability, motivation or opportunity regardless of design quality. This conditional effectiveness dynamic is the central theoretical gap that the review addresses through the contextually extended AMO framework.

A growing body of critical scholarship argues that dominant AI development paradigms embed Global North epistemological assumptions about individual agency, institutional stability, labour market formalisation and technological rationality that are not universally applicable and may be actively distorting when deployed in Global South contexts (Birhane, 2021; Mohamed et al., 2020). These scholars draw on decolonial theory to argue that the dominant AI development ecosystem, from training data composition and algorithmic design to deployment governance and impact assessment, reflects and reproduces the values, priorities and power structures of North American and European technology centres, marginalising African and other developing-world knowledge systems.

For GHRM in SSA, this decolonial critique carries direct and consequential implications. AI recruitment tools trained predominantly on North American or European hiring data may embed cultural biases, linguistic assumptions and socioeconomic proxies, such as educational institution prestige, professional network composition and digital communication style, that systematically disadvantage qualified candidates from African professional and educational backgrounds (Raghavan et al., 2020). A decolonial approach demands more than contextual adaptation of existing tools; it requires participatory reconstruction of AI-GHRM frameworks through sustained engagement with SSA scholars, practitioners, workers and community stakeholders as co-designers.

This study uses a systematic literature review (SLR) methodology, selected for its capacity to minimise selection bias, ensure reproducibility and provide a transparent and auditable account of the evidence base (Tranfield, Denyer, and Smart, 2003; Kitchenham and Charters, 2007). Unlike narrative or traditional literature reviews, which are susceptible to subjective article selection and interpretive bias, the SLR follows a pre-defined protocol that governs each stage of the review process, from research question formulation through database search, screening, quality appraisal, data extraction and thematic synthesis. The review is guided by the PRISMA 2020 protocol (Page et al., 2021), widely adopted as the reporting standard for systematic reviews across management and social science disciplines and uses the three-stage thematic synthesis approach proposed by Thomas and Harden (2008): inductive free line-by-line coding of study findings, organisation of codes into descriptive themes and development of analytical themes that offer interpretive insights into the research questions.

Systematic searches were conducted across five major electronic databases: Scopus, Web of Science (WoS), EBSCO Business Source Complete, ProQuest Dissertations and Theses and Google Scholar. These databases were selected for their comprehensive coverage of management, HRM, sustainability and information systems literatures and for their indexing standards that ensure inclusion of high-quality peer-reviewed scholarship. Google Scholar was used as a supplementary source specifically to enhance coverage of African-authored scholarship not consistently indexed in primary databases. The search was conducted between January and April 2025, covering publications from January 2018 to December 2025, a temporal boundary reflecting the emergence of AI-enabled HRM as a distinct research domain. Forward citation tracking of seminal articles and ancestry searching of reference lists supplemented the database search. The full search strategy is presented in Table 1.

All retrieved records were assessed against pre-defined inclusion and exclusion criteria, developed prior to the search to minimise post hoc rationalisation of study selection decisions (Moher et al, 2009). Both empirical and conceptual studies were included, reflecting the nascent character of AI-enabled GHRM as a field in SSA, where theoretical contributions play a critical role in shaping the research agenda. The objective was not to obtain a statistically representative sample but to comprehensively identify all relevant peer-reviewed studies that met the eligibility criteria. This approach is consistent with established systematic literature review methodology, which prioritises transparency, reproducibility and methodological rigour (Tranfield et al., 2003; Page et al., 2021).

Disagreements were resolved through discussion and reference to the original source texts. The full criteria are presented in Table 2.

The study selection process followed the four-phase PRISMA flow: identification, screening, eligibility and inclusion (Page et al., 2021). All records retrieved from database searches were imported into Rayyan, a Web-based systematic review management platform, which facilitated duplicate detection and collaborative blind screening. Of 1,214 initially identified records, 268 duplicates were removed, yielding 946 records for title and abstract screening. Following this screening phase, 215 full-text articles were retrieved and assessed for eligibility against the inclusion and exclusion criteria. A further 157 were excluded at the full-text stage for insufficient construct focus (n = 82), exclusive Global North orientation without comparative relevance (n = 48), failure to meet quality appraisal standards (n = 18) or inaccessibility of full text (n = 9). A final corpus of 58 peer-reviewed studies was included in the thematic synthesis. The complete PRISMA summary is presented in Table 3.

Quality appraisal used an adapted MMAT/CASP instrument (Hong et al., 2018), scoring studies across five dimensions: methodological rigour, theoretical grounding, contextual specificity, validity and reliability and contribution to synthesis. The average quality score of included studies was 19.1 out of 25.

A standardised data extraction form captured bibliographic details, research design, theoretical frameworks used, key constructs and variables, principal findings relating to each research question, identified limitations and contextual conditions. Data extraction was performed by the primary reviewer, with a second reviewer performing a double extraction; discrepancies were resolved through discussion. NVivo 14 qualitative data analysis software was used to manage the coding process, facilitate audit trail documentation and support the identification of thematic patterns across the corpus. Inductive coding progressed iteratively through descriptive to analytical themes, with particular attention to the contextual specificity of findings in SSA settings. Analytical themes were developed by examining patterns of convergence, divergence and productive tension across the coding categories, organised around the two RQs.

Thematic analysis of the 58 included studies identified four dominant themes, presented in Table 4. Studies published between 2022 and 2025 account for 56.9% (n = 33) of the corpus, indicating accelerating scholarly attention to the focal intersection. Geographically, South Africa (n = 18) and Nigeria (n = 11) contribute the largest national shares, with East African economies (Kenya, Tanzania, Ethiopia; n = 7 collectively), West African economies outside Nigeria (n = 5) and Central African economies (n = 3) substantially underrepresented. This geographic skew reflects both the relative maturity of SSA’s largest economies as research sites and a systematic gap in the evidence base for less digitally developed national contexts. Sectorally, manufacturing (34.5%) and services encompassing hospitality, banking and retail (29.3%) are the most frequently studied sectors, while agriculture, extractives and the informal economy, which together dominate SSA employment, account for fewer than 15% of corpus studies. Thematic findings and study distribution are shown in Table 4.

AI-powered applicant tracking systems (ATS) enable organisations to systematically screen candidates not only for technical competencies but also for environmental value alignment, sustainability-related experiences and pro-environmental behavioural dispositions identified through psychometric modelling and NLP of application materials (Ren et al., 2021; van Esch and Black, 2019). Within the SSA corpus, green recruitment represents the most consistently evidenced AI-GHRM enabling mechanism. Studies from South African financial services organisations (n = 5) and Nigerian manufacturing firms (n = 4) demonstrate measurable improvements in the environmental fit of recruited cohorts following AI-assisted screening implementation, with reductions in time-to-hire and expansion of talent pool geographic reach as additional reported benefits.

The compensatory role of mobile learning platforms in bridging infrastructure access gaps emerges as a particularly significant and SSA-distinctive finding. Studies consistently demonstrate that organisations deploying mobile-based, AI-personalised LMS achieve substantially broader green training programme coverage in terms of geographic reach, employee participation rates and learning continuity than those dependent on fixed desktop infrastructure (GSMA, 2023; ITU, 2023). This finding is consistent with SSA’s documented pattern of mobile internet leapfrogging: the continent’s mobile broadband adoption has substantially outpaced fixed-line infrastructure development, creating a de facto mobile-first digital environment in which mobile platforms are not a compromise but the primary mode of digital engagement for large segments of the SSA workforce. These findings position mobile technology as the strategically appropriate primary delivery vehicle for green training in SSA rather than a compromise channel.

Real-time environmental dashboards and AI-driven continuous feedback systems improve the consistency, transparency and motivational salience of green performance management in organisational contexts with adequate digital infrastructure (Tambe et al., 2019). Automated environmental KPI tracking, machine learning-based anomaly detection in resource consumption and predictive performance modelling enable HR managers to identify environmental underperformance early and intervene proactively, transforming green performance management from a periodic appraisal exercise into a continuous, data-driven governance process. Within the SSA corpus, evidence for this mechanism is more contextually qualified than for green recruitment or mobile training. Studies from larger South African and Kenyan corporate contexts report meaningful improvements in environmental KPI adherence and green behaviour consistency following AI performance monitoring implementation. By contrast, studies from smaller enterprises and less connected settings across West and Central Africa document implementation failures attributed to connectivity instability that renders dashboard systems unreliable, HR professional interpretive capacity gaps that prevent meaningful use of AI-generated performance insights and employee resistance to algorithmic surveillance perceived as intrusive or inequitable. This conditional effectiveness pattern underscores the centrality of digital readiness assessment as a prerequisite for AI performance system deployment.

AI capabilities for cross-cutting resource optimisation encompassing machine learning-driven energy management systems, AI-facilitated paperless HR process transformation and smart building integration for workplace environmental monitoring constitute a theoretically significant but empirically underdeveloped enabling mechanism in the SSA corpus (Hossain et al., 2020). Of 38 studies contributing to the RQ1 theme, fewer than ten address resource optimisation specifically, and those that do predominantly use conceptual or secondary data analytical designs rather than primary empirical evidence. Primary empirical evidence of AI-GHRM-attributable environmental footprint reductions in SSA organisations is notably absent from the corpus reflecting both the early stage of AI-GHRM deployment in the region and the methodological challenges of isolating HR-specific AI contributions to environmental outcomes.

Intersecting with infrastructure constraints, digital skills gaps among HR professionals encompassing AI literacy, data analytics capability, systems integration competence and digital change management skills are identified in 38 studies as co-determining limitations (Charlwood and Guenole, 2022). Without investment in these capabilities, SSA organisations risk acquiring AI-GHRM tools that they lack the organisational capacity to implement effectively, interpret responsibly or adapt contextually. Institutional voids encompassing weak or unenforced data protection legislation, limited professional HR associational infrastructure, inconsistent environmental regulation and minimal state AI governance capacity further reduce both the normative pressures to adopt GHRM and the governance conditions needed to deploy AI responsibly and equitably in HR contexts (Wirtz, Weyerer and Geyer, 2019). Critically, these three constraint dimensions are mutually reinforcing rather than additive. Organisations that invest in AI-GHRM platforms without addressing capability gaps find technology rendered ineffective. Those developing digital HR literacy without a reliable infrastructure find capabilities stranded. Those in institutional voids face perverse incentives discouraging sustainability investment. This interconnected constraint structure demands systems-level policy and organisational responses addressing all three dimensions simultaneously, rather than piecemeal technological or training interventions.

The corpus documents cases in which AI recruitment tools trained on Global North data sets produced systematic disadvantage for candidates from non-Western cultural and linguistic backgrounds (Raghavan et al., 2020) and in which AI performance monitoring systems calibrated to Global North productivity norms misrepresented the contributions of workers in communally organised or informally structured work settings prevalent across SSA (Birhane, 2021). The decolonial critique of Mohamed et al. (2020) is directly applicable in AI-GHRM tools developed without the participatory engagement of SSA scholars, practitioners and workers, who risk reproducing and amplifying existing structural inequalities under the guise of algorithmic objectivity.

The practical stakes of algorithmic bias in SSA GHRM contexts are substantial. AI recruitment tools that systematically disadvantage candidates from marginalised communities violate employment equity principles and erode organisational social license, a particularly acute concern in SSA contexts characterised by high unemployment and significant social inequality. AI performance monitoring tools that misrepresent contributions create perceptions of procedural unfairness that, according to the social exchange theory (Latan et al., 2018), will reduce reciprocal green organisational commitment and undermine precisely the pro-environmental behavioural outcomes that GHRM seeks to cultivate. Algorithmic bias management is therefore not a peripheral compliance concern but a central condition for the ethical and effective implementation of AI-enabled GHRM in SSA.

This internal heterogeneity has significant theoretical implications. It suggests that contextual moderator models for AI-enabled GHRM must be calibrated to specific national and sectoral digital readiness profiles, incorporating dimensions such as broadband penetration, development of the cloud services market, AI skills supply, data protection legislative maturity and strength of HR professional associations, rather than to a generic SSA ideal type. It also implies that the review’s findings, weighted toward South African and Nigerian evidence, may substantially overstate what is achievable in less digitally developed SSA contexts, and that the design of AI-GHRM interventions must be differentiated across the continent’s diverse institutional landscape.

The most substantive theoretical contribution of this review is the contextually extended AMO framework for AI-enabled GHRM in SSA. Building on Ren et al.’s (2021) extension of the AMO model (Appelbaum et al., 2000; Renwick et al., 2013), the review identifies three threshold boundary conditions that determine whether AI amplifies or fails to amplify each AMO dimension in SSA organisational contexts: digital infrastructure adequacy, organisational AI capability and institutional governance quality. The review proposes that these conditions function as threshold moderators rather than continuous moderators, that is, that AI-AMO-GHRM relationships manifest and strengthen only above certain minimum infrastructure and capability thresholds, below which AI deployment produces negligible or negative AMO effects regardless of tool design quality. This threshold moderator specification is an empirically testable theoretical proposition that advances the AMO literature beyond its implicit Global North assumptions and provides a theoretically grounded basis for predicting when and where AI-enabled GHRM will be effective in developing-world settings.

The proposal to treat digital ecosystem maturity as a fourth institutional pillar extends the institutional theory (DiMaggio and Powell, 1983) in a theoretically productive and practically significant direction. The conventional tripartite framework of regulative, normative and cognitive institutional pressures does not explicitly account for digital ecosystem conditions, broadband availability, cloud service market development, AI skills supply, digital governance architecture and technology standard diffusion as institutional shapers of organisational AI adoption capacity. The SSA corpus consistently identifies these conditions as powerful determinants of AI-GHRM adoption, operating independently of and in complex interaction with conventional institutional pressures. This extension carries applicability beyond GHRM to the broader sociology of digital transformation in developing economies, offering a theoretical resource for scholars examining how institutional environments shape organisational responses to the AI transition across diverse global contexts.

The integration of decolonial AI theory into GHRM discourse constitutes the review’s most novel theoretical contribution. Drawing on Birhane (2021) and Mohamed et al. (2020), the review foregrounds epistemic justice, participatory design and algorithmic accountability within the GHRM framework, opening a line of inquiry not previously addressed in management scholarship. The decolonial perspective challenges the epistemological foundations of dominant AI-GHRM frameworks, demanding a reorientation from technology transfer to co-creation and from efficiency optimisation to equity and justice, with direct implications for research design, technology development and practice.

For SSA organisations and HR managers, the review’s findings support a phased, capability-sensitive AI-GHRM implementation model calibrated to the specific digital readiness profile of the organisation and its operating environment. Organisations in digitally constrained environments should prioritise interventions with lower infrastructure thresholds, such as mobile-based green training, basic digital HR record-keeping and simple environmental KPI reporting dashboards, before investing in more sophisticated AI applications such as predictive performance modelling, AI-driven recruitment analytics or enterprise-wide resource optimisation systems. The review further advises that a digital readiness assessment, evaluating infrastructure reliability, HR AI capability and institutional governance adequacy, should be conducted as a systematic prerequisite for AI-GHRM investment decisions.

For governments and policymakers across SSA, the review establishes that the enabling conditions for AI-enabled GHRM, reliable digital infrastructure, data protection regulation, AI governance frameworks and credibly enforced environmental compliance regimes are public goods that individual organisations cannot provide independently. Regional coordination mechanisms, including data governance harmonisation and cross-border digital infrastructure investment, offer opportunities to accelerate progress beyond what is achievable at national level alone.

For AI technology developers, the review underscores both the ethical and commercial imperative to design AI-GHRM tools appropriate for SSA deployment. This requires developing low-bandwidth, mobile-optimised platforms; training AI models on culturally and linguistically diverse African data sets; conducting algorithmic bias audits across SSA population groups before deployment; and adopting participatory design methodologies that engage SSA practitioners, workers and community stakeholders from earliest development stages. SSA’s rapidly expanding formal economy and growing corporate sustainability agenda represent significant, largely untapped markets for contextually grounded AI-GHRM solutions. Also, to ensure that AI-enabled GHRM systems are socially inclusive and contextually relevant, AI developers should move beyond technology-driven implementation towards participatory and context-sensitive design approaches. Consistent with a decolonial AI perspective, the development of AI applications should actively involve local stakeholders, including HR practitioners, employees, policymakers, labour representatives and community organisations from SSA throughout the design, implementation and evaluation processes. Such engagement helps ensure that AI systems reflect local organisational realities, cultural values and sustainability priorities rather than reproducing assumptions embedded in technologies developed for high-income contexts.

The framework in Figure 1 integrates five theoretical layers: AI and digital transformation enablers (Layer 1, purple), the extended AMO model (Layer 2, teal), GHRM practices embedded within the SSA contextual moderator boundary (Layer 3), sustainability outcomes (Layer 4, amber) and the decolonial and critical cross-cutting layer (Layer 5, grey). Dashed borders denote contextual boundary conditions, digital infrastructure adequacy, organisational AI capability and institutional governance quality that function as threshold moderators of inter-layer relationships in Sub-Saharan African organisational contexts. The digital ecosystem maturity pillar (right moderator column, Layer 3) extends DiMaggio and Powell’s (1983) institutional theory as a fourth institutional dimension. The decolonial layer (Layer 5) integrates the critical frameworks of Birhane (2021) and Mohamed, Png and Isaac (2020). OCB-E = organisational citizenship behaviour environment; SDGs = Sustainable Development Goals.

The restriction to English-language publications introduces a systematic bias against scholarship from Francophone and Lusophone African contexts encompassing Côte d‘Ivoire, Cameroon, Senegal, the DRC, Mozambique and Angola, among others, that may host distinctive AI-GHRM dynamics not captured in the English-language corpus. The geographic concentration of available studies in South Africa and Nigeria limits the generalisability of findings to Central, East and West African economies where empirical AI-GHRM research is almost entirely absent; findings from SSA’s most digitally advanced economies may substantially overstate what is currently achievable in less developed national contexts. Consequently, future comparative research should extend this work to Francophone and Lusophone Sub-Saharan African contexts, where distinct legal traditions, governance systems, educational structures and institutional environments may shape the adoption and implementation of AI-enabled GHRM differently. Such comparative analyses would provide a more inclusive understanding of digital transformation across the region and contribute to the development of contextually grounded and decolonised AI governance frameworks.

This systematic literature review has established that AI and digital transformation technologies hold genuine and substantive potential for advancing GHRM in Sub-Saharan African organisations, enhancing the reach, intelligence and environmental impact of green recruitment, training, performance management and resource optimisation across the region’s diverse and rapidly evolving organisational landscape. This potential, however, is contingent, contextually bounded and ethically complex rather than automatic or universal. Theoretically, the review’s contextually extended AMO framework specifying threshold boundary conditions on AI-AMO-GHRM relationships, the digital ecosystem institutional pillar extending the institutional theory and the integration of decolonial AI perspectives together provide a substantially more contextually sensitive and empirically grounded theoretical apparatus than the current Global North-anchored literature offers. These extensions are not merely additive refinements but represent a qualitative reorientation of the field towards contextual authenticity, epistemic humility and equitable engagement with the communities and organisations that stand to benefit most from sustainable AI-enabled HRM.

The future research agenda advanced by this review encompasses six priority directions. Firstly, longitudinal research designs, including panel surveys, experience sampling methodology and longitudinal case studies, are urgently needed to assess whether AI-enabled GHRM interventions produce durable improvements in green employee behaviour and organisational environmental performance over 12- to-36-month periods, or whether observed cross-sectional effects reflect short-term compliance rather than sustained behavioural change. Secondly, comparative cross-national studies within SSA examining how different national configurations of digital infrastructure, institutional development and organisational capability moderate AI-GHRM effectiveness would provide the contextually differentiated evidence base that the current literature lacks and that policymakers urgently need. Thirdly, while this review develops an integrative conceptual framework linking AI, digital transformation and GHRM in the Global South, future research should move beyond conceptual development towards empirical validation. Specifically, the framework proposes that the effectiveness of AI-enabled GHRM is contingent upon three threshold boundary conditions: digital infrastructure adequacy, organisational AI capability and institutional governance quality. Operationalising these constructs will facilitate hypothesis testing and strengthen theory development.