AI-driven knowledge management for sustainable businesses: a comprehensive analysis Open Access

The accelerating convergence of environmental, social and economic pressures has propelled sustainability to the forefront of organizational strategy. Firms face mounting demands to optimize resource use, reduce carbon footprints and innovate circular business models while navigating complex regulatory landscapes (Khan et al., 2024; Di Vaio et al., 2020). Knowledge management (KM), defined as the systematic creation, sharing and application of organizational knowledge, has long been recognized as a critical capability for achieving sustainable performance (Gupta et al., 2022; Scarso and Bolisani, 2024). However, traditional KM approaches often struggle with the vast data streams and dynamic contexts inherent in sustainability challenges.

In response, artificial intelligence (AI) has emerged as a transformative enabler within knowledge management systems (KMS) (Nakash and Bolisani, 2025). AI technologies (e.g. machine learning and natural language processing [NLP]) facilitate real-time decision-making by extracting insights from unstructured data (Simon et al., 2022). Complementary tools like big data analytics (BDA) and knowledge graphs further support knowledge structuring and contextualization (Yu, 2024). For instance, predictive analytics can forecast resource demand and emissions trajectories, while AI-driven systems optimize supply chains for minimal environmental impact (Baumont De Oliveira et al., 2021; Poch et al., 2004). Moreover, these advances align with the emerging Industry 5.0 paradigm, which shifts the focus from pure automation toward human-centricity, sustainability and resilience, prioritizing human–machine synergy to democratize technology and enhance stakeholder collaboration (Gronau and Grum, 2024; Takahashi et al., 2024). Together, they promise to bridge the gap between KM’s theoretical potential and sustainability’s practical demands.

Despite growing scholarly attention, the specific integration of AI and KM for addressing sustainability remains underexplored. Several systematic reviews have examined AI–KM intersections generally. For instance, recent studies have synthesized people-technology synergies in KMS (Pai et al., 2022), explored AI impacts on KM practices (Taherdoost and Madanchian, 2023), developed roadmaps for AI-enabled KM (Nakash and Bolisani, 2024), surveyed AI applications in business KM (Thakuri et al., 2024) and analyzed AI implementation strategies in KM (Novalin et al., 2024). However, these works rarely incorporate sustainability as a central focus. A few studies have begun exploring this nexus: Di Vaio et al. (2020) explicitly frame AI-driven KM within the UN sustainable development goals (SDGs). More recently, Sorout and Singh (2025) conducted a bibliometric analysis of 66 Scopus documents, mapping collaborative networks to illustrate how AI-enhanced KMS broadly support environmental, social and economic sustainability. Additionally, Cai et al. (2025) propose a conceptual framework linking AI-driven KM with startup sustainability.

While these prior reviews provide valuable foundational insights, critical research gaps remain. Theoretically, existing reviews often lack a unifying framework to conceptualize how AI and KM foster dynamic capabilities (DC), organizational creativity and sustainable performance (Almheiri et al., 2025; Cai et al., 2026). Methodologically, they predominantly rely on single databases, and use purely quantitative bibliometrics without deep qualitative abstraction (Sorout and Singh, 2025). Furthermore, while the field is shifting from techno-centric automation toward human-centric AI–KM models (e.g. Lean Service 5.0), this transition is frequently hindered by sociocultural inhibitors. These human-level barriers, such as cultural resistance and lack of trust, are rarely addressed in systematic reviews, particularly within complex ecosystems like value co-creation (Bonamigo et al., 2025b). To address these gaps, the originality of this study lies in its comprehensive, mixed-methods approach that integrates quantitative bibliometric profiling of dual databases (Scopus and Web of Science [WoS]) with an in-depth qualitative thematic content analysis of 80 selected articles. This enables the construction of a comprehensive tripartite theoretical framework mapping the technological foundations, organizational strategies and ecological outcomes of AI-driven KM, ultimately positioning human-centric principles at the core of sustainable development. Guided by the following three research questions (RQs), this study systematically synthesizes the literature up to 2025 through the lens of the knowledge-based view (KBV):

This study rests on three distinct yet increasingly intersecting domains: KM, AI and sustainability. Building on its foundational definition, KM is theoretically grounded in the KBV, which positions intellectual assets as a firm’s most critical strategic resource (Grant, 1996). In this context, KM functions as a socially mediated infrastructure facilitating the dynamic conversion of tacit and explicit knowledge (Alavi and Leidner, 2001), ultimately translating raw data into strategic insights (Nonaka, 1994). Within management and business literature, AI is conceptualized as a suite of general-purpose technologies enabling systems to interpret external data, learn from it and achieve specific goals through flexible adaptation (Kaplan and Haenlein, 2019), drastically reducing prediction costs in complex environments (Agrawal et al., 2022). Sustainability, anchored in the Brundtland Commission’s definition [World Commission on Environment and Development (WCED), 1987], is operationalized in organizational research through the triple bottom line framework (Elkington, 1998), mandating the simultaneous optimization of environmental, social and economic performance.

To establish a cohesive theoretical rationale for integrating AI and KM to achieve sustainability, this study positions the KBV (emphasizing knowledge as the core strategic asset) as its primary analytical anchor, supplemented by DC theory (addressing organizational adaptation) and socio-technical systems (STS) theory (highlighting the alignment of technology and people).

From the primary perspective of the KBV (Grant, 1996), AI’s immense computational power and the massive data sets it processes are not inherently sources of sustained competitive advantage. Instead, KM provides the critical strategic architecture required to structure AI-generated insights into codified organizational routines and green intellectual capital. To explain how these knowledge assets are operationalized in highly volatile environments, we draw upon the DC framework as a supplementary mechanism (Teece, 2007). Sustainability challenges demand that firms possess the capacity to sense environmental shifts, seize new opportunities and reconfigure their asset base. Here, AI and KM play complementary roles in forging a “Green Dynamic Capability.” AI serves as the ultimate sensing mechanism, forecasting environmental trajectories with extreme precision. However, because AI alone lacks contextual judgment, the KBV-driven KM infrastructure provides the crucial seizing and reconfiguring mechanisms to translate these technical forecasts into strategic actions.

Furthermore, while KBV and DC explain the strategic conversion of knowledge, the STS perspective (Cherns, 1976) explains its necessary social embedding. Effective digital implementation requires the joint optimization of technical subsystems (AI algorithms) and social subsystems (human employees, culture and ethics). KM bridges this gap by establishing the participatory routines (e.g. continuous feedback loops and shared vocabularies) necessary to translate opaque algorithmic outputs into trusted, actionable knowledge. This actively facilitates human–AI (HAI) collaboration, combining experiential knowledge and human creativity with machine efficiency to mitigate technostress and foster continuous learning.

In synthesis, AI provides the analytical velocity required to tackle complex ecological problems, while KM provides the organizational structure and human-centric governance necessary to translate those predictions into sustainable actions. Together, they transform static environmental data into green intellectual capital, enabling organizations to innovate circular business models and achieve enduring sustainable development.

This study adopts a mixed-methods framework that integrates quantitative bibliometric analysis with qualitative thematic content analysis. Following systematic literature review protocols (Creswell and Clark, 2017; Venkatesh et al., 2016) and the procedures outlined by Denyer and Tranfield (2009) and Kitchenham (2004), the research proceeded in three phases: (1) data collection, (2) bibliometric profiling and science mapping and (3) thematic content analysis. This mixed-methods approach allows for both a macro-level overview of research dynamics and an in-depth interpretation of underlying theoretical mechanisms.

Literature was retrieved from Scopus and WoS due to their multidisciplinary coverage and citation reliability (Martín-Martín et al., 2021). The optimized search string targeted three focal constructs: (“knowledge management” OR “KM” OR “knowledge processes” OR “knowledge systems”) AND (“artificial intelligence” OR “AI” OR “machine learning” OR “deep learning” OR “data-driven” OR “big data” OR “data analytics”) AND (“sustainab*”). The asterisk (*) served as a wildcard to capture variations like “sustainable.” The search was limited to titles, abstracts and keywords. The initial data retrieval was conducted in early 2025. To ensure the review captured the most cutting-edge advancements, a comprehensive update was performed on January 6, 2026, covering all literature published up to December 31, 2025. We followed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines (Page et al., 2021) to ensure methodological rigor (Figure 1).

The initial search yielded 1,932 records (1,157 from Scopus and 775 from WoS). Before screening, 712 records were removed, including 442 duplicates and 270 items excluded based on document type or language. The remaining 1,220 records underwent title, abstract and keyword screening, leading to the exclusion of 1,029 nonrelevant articles because their primary focus fell outside the substantive scope of the review. Of the 191 reports sought for retrieval, 13 full texts were unavailable. The remaining 178 articles advanced to full-text review. Following exclusions for scope limitations (n = 49) and partial relevance (n = 50), 80 studies were ultimately selected for analysis, comprising 79 articles from the database search and one identified through citation tracking.

To profile the domain’s evolution, we generated descriptive bibliometrics, including annual publications, citation counts and methodological patterns (Donthu et al., 2021; Gupta and Bhattacharya, 2004). For deeper spatial and network insights, we performed advanced science mapping in RStudio using the Bibliometrix R-package (Aria and Cuccurullo, 2017). Specifically, Bibliometrix routines mapped the geographic distribution, constructed a keyword co-occurrence network and generated a thematic map to visualize the intellectual structure and thematic clusters.

Qualitative analysis of the 80 articles followed a three-step procedure (Elo and Kyngäs, 2008): open coding, categorization and abstraction. Open coding extracted passages on AI techniques, KM processes, sustainability objectives, enablers, barriers and outcomes. Categorization grouped codes by technological focus (e.g. green analytics and decision-support), theoretical contributions (e.g. maturity models and DC), application domain (e.g. manufacturing and education), human and social factors (e.g. stakeholder collaboration) and practical outcomes (e.g. eco-innovation and performance metrics). Abstraction synthesized six overarching themes that reflect the field’s landscape: green knowledge management (GKM) and data-driven innovation; AI and big data integration with KM; KM frameworks and maturity models; decision-support systems (DSS) and platforms; human-centric KM and socialization; and industry-specific KM applications. To provide a coherent intellectual structure, these themes were organized into three pillars: technological foundations, organizational strategy and human agency and ecological outcomes. Team discussions ensured coding consensus and thematic reliability.

Research on AI–KM–sustainability integration spans over two decades. Between 2004 and 2025, 80 articles were published, with the vast majority (75 articles, representing nearly 94% of the total) appearing between 2019 and 2025. Figure 2 illustrates this growth trajectory. The early phase (2004–2016) featured nascent and scattered research. However, the field has entered a phase of exponential expansion since 2017, demonstrating a robust compound annual growth rate of approximately 52%. Most notably, a significant inflection point occurred recently: publications nearly doubled from 8 in 2023 to 15 in 2024, and surged to 28 in 2025. This remarkable 87% year-over-year increase indicates that the intersection of AI, KM and sustainability has transitioned from a niche topic to a rapidly intensifying research priority.

The 80 selected articles had accumulated a total of 2,625 citations by January 2026 (Figure 3). Although annual patterns fluctuate, the distribution highlights distinct waves of intellectual influence. The early period is anchored by 2004, where the foundational work of Poch et al. (2004) alone contributed 239 citations (almost 10% of the total). Following a period of relative dormancy between 2012 and 2019, the field experienced a resurgence. The most significant impact was observed in 2020, reaching a notable peak of 989 citations. This surge is predominantly attributed to the seminal literature review by Di Vaio et al. (2020), which generated 870 citations and established the field’s core agenda regarding AI and SDGs. Notably, recent research has gained immediate traction: papers published in 2024 have already amassed 378 citations, and the newly published 2025 cohort has quickly gathered 133 citations. Overall, approximately 89% of all citations (2,333 out of 2,625) are attributed to research published from 2019 onwards, underscoring that the domain is not only growing in volume but also in scholarly impact.

The 80 selected articles involved contributing authors from 39 countries and regions across five continents (Figure 4). The numbers represent total publications by the institutional regions of the authors. Geographically, Asia dominates the research landscape with 90 institutional contributions, significantly outpacing Europe (34), the Americas (11), Oceania (2) and Africa (1). Within Asia, China (15) and India (13) emerge as the primary global research hubs. Significant contributions also originate from Malaysia and Saudi Arabia (eight each), followed by active clusters in Iraq and the UAE (six each). In Europe, research activity is led jointly by France and Italy (seven each), with steady contributions from Germany (4), the UK (3) and Cyprus (3). The Americas are primarily represented by the USA (8), alongside Canada (2) and Colombia (1). Oceania is represented solely by Australia (2), while Libya (1) provides the sole contribution from Africa. Regarding collaboration patterns, 48 articles involved authors from a single region, while 32 represented international or inter-regional collaborations.

As illustrated in Figure 5, quantitative approaches represent the dominant methodological paradigm, accounting for 40 studies (50%). Within this category, 33 articles (41%) are empirical quantitative studies employing statistical techniques (e.g. partial least squares structural equation modeling and machine learning models) primarily via online or multi-mode surveys. The remaining seven quantitative studies (9%) follow a design-oriented approach, focusing on model or algorithm development via experiments and simulations. Mixed-methods designs constitute the second largest category with 17 articles (21%), reflecting a growing trend to combine quantitative rigor with qualitative depth. Conceptual papers account for 15 studies (19%), advancing theoretical frameworks. Qualitative research is represented by seven studies (9%), comprising case studies, in-depth interviews and advanced corpus-based discourse analysis using NLP. Additionally, one study published in 2025 is classified as a pure literature review.

To elucidate the intellectual structure of the field, we extracted keywords from the 80 articles. Following data standardization (e.g. merging synonymous terms), we selected the 28 terms appearing at least twice. Using these high-frequency keywords, we constructed an undirected co-occurrence network using the igraph and ggraph packages in RStudio. The resulting network is linked by 102 edges, each representing a paired appearance of two keywords (Figure 6). Following Radhakrishnan et al. (2017), the network density is 0.27, and the average node degree is 7.29, meaning each keyword co-occurs with approximately seven others.

Five primary thematic hubs anchor this polycentric network:AI (Degree 22), KM (Degree 21), BDA (Degree 14), sustainability (Degree 12) and KBV (Degree 11). Edge weight analysis reveals robust connections among this core group. For instance, AI is most strongly linked with KM (Weight 13) and sustainability (Weight 9), while KM shows significant ties to sustainable development (Weight 6).

The Louvain community detection algorithm (resolution = 0.85) reveals three coherent subthemes:

• Digital and conceptual foundation (Blue cluster): Serving AI, KM and sustainability at its core, this infrastructure layer highlights the macro-level convergence of these pillars, supported by foundational enablers such as machine learning, big data and Industry 4.0.

• Strategic management perspective (Green cluster): Rooted in the KBV, this cluster demonstrates how organizations leverage GKM and green intellectual capital to build DC and achieve sustainable performance.

• Action-to-outcome pathway (Orange cluster): This stream reveals a clear, data-driven operational sequence, illustrating that specific data tools (e.g. BDA and KMS) are closely coupled with tangible environmental outcomes like green innovation and the circular economy.

For thematic clustering, we mapped these keyword clusters onto a strategic diagram based on Callon’s centrality (external connectivity) and density (internal cohesion). Figure 7 visualizes these macro-topics, identifying their strategic maturity and suggesting clear directions for future research (Ramos Cordeiro et al., 2024).

• Motor theme (Q1): The core driving cluster (AI, KM and sustainability) exhibited the highest density (1.698) and strong centrality (1.689). This confirms that AI and KM convergence serves as the strategic conceptual backbone for sustainability, representing a well-developed hub that drives the domain’s intellectual structure.

• Niche theme (Q2): The strategic management cluster (KBV and GKM) displayed positive internal cohesion (0.323) but negative external connectivity (−0.160). This indicates a mature yet isolated specialized research stream, suggesting an opportunity for future cross-disciplinary integration.

• Emerging theme (Q3): The ecological applications cluster (green innovation, circular economy) showed low density (−0.323) and low centrality (−0.618). This reveals a significant gap: while specific environmental applications are gaining traction, they are currently in the early stages of integration with broader AI and KM literature.

• Basic theme (Q4): The technological foundation cluster (big data and machine learning) combines positive external connectivity (0.160) with lower internal cohesion (−0.511). These transversal building blocks are broadly applied as fundamental tools, yet their specific internal theoretical frameworks regarding sustainability remain under development.

Building on the bibliometric structure identified in Section 4, a rigorous thematic content analysis of the 80 selected articles uncovers substantive research streams. Following the abstraction procedure (Section 3.3), we synthesized the bibliometric findings into six overarching themes organized under three strategic pillars: (1) technological foundation, (2) organizational strategy and human agency and (3) ecological outcomes. Table 1 presents these themes, illustrating how KM, AI and data-driven approaches converge to advance sustainable practices.

The integration of AI and BDA into KMS represents a fundamental shift from static knowledge repositories to dynamic, predictive ecosystems. This convergence creates a “digital synergy” where AI capabilities act as a critical antecedent for enhancing organizational learning agility and DC, which are essential for navigating sustainable development.

Specifically, deploying AI capabilities empirically enhances DC, allowing firms to sense environmental changes and reconfigure resources more effectively. For instance, AI-driven circular transformation is conceptualized as a strategic capability that fosters ecosystem collaboration and resilience in startups navigating circular economy demands (Hwang et al., 2025). In the context of corporate environmental sustainability, AI technology innovation serves as a primary driver, but its impact is significantly mediated by KM capacity, suggesting that technological tools require robust knowledge processes to realize their green potential (Zu et al., 2025). Furthermore, advanced computational techniques such as hybrid SEM-ANN models reveal that KM factors (Al-Sharafi et al., 2023), particularly knowledge acquisition and application, are the most critical predictors for the sustainable use of generative AI (GenAI) across organizational settings (Al-Emran et al., 2025; Al-Qaysi et al., 2025). Sector-specific evidence further illustrates this: coupling BDA with agile supply chains and KM drives sustainable product innovation, highlighting how data velocity and variety accelerate green research and development cycles (Adiguzel et al., 2025; Mohammad et al., 2025). Moreover, novel paradigms like transfer learning are emerging as sustainable solutions, enabling knowledge models to be reused across different domains to conserve computational resources and energy (Gurjar and Voditel, 2022).

DSS for sustainability are evolving from isolated tools into decentralized platforms that leverage emerging technologies to ensure trust, transparency and multi-stakeholder collaboration. Functioning as the basic, foundational building blocks, these systems primarily reduce “knowledge asymmetry” and manage the high volume of heterogeneous data required for environmental governance and supply chain traceability. This evolution marks a transition toward ecosystems where data integrity and real-time analytics drive credible sustainability management.

Recent advancements highlight blockchain’s critical role in securing the knowledge lifecycle. By creating immutable ledgers, blockchain-based frameworks ensure environmental data remains tamper-proof and transparent, which is vital for building trust among supply chain partners (Sheela and Rajini, 2023; Singh et al., 2025). Integrated with AI, these systems analyze complex knowledge flows to predict risks and optimize real-time resource allocation. Parallelly, knowledge graphs effectively manage the unstructured data inherent in product lifecycles. For example, graph-based approaches assist in the evolutionary design of smart product-service systems, enabling designers to navigate vast user-generated data to create sustainable, personalized solutions (Li et al., 2020). In domains like hydrology and vertical farming, AI-driven DSS platforms bridge educational knowledge practices and environmental monitoring, utilizing sensor data to foster ecosystem resilience (Baumont De Oliveira et al., 2021; Gurler et al., 2025). Furthermore, manufacturing “zero-defect” platforms serve as collaborative architectures connecting technology providers with manufacturers. By algorithmically matching AI methodologies to specific production needs, these platforms facilitate seamless knowledge exchange, allowing for proactive defect prediction and waste reduction to align operational efficiency with sustainability metrics (Berndt et al., 2024).

To effectively operationalize AI and KM for sustainability, organizations require robust strategic frameworks and maturity models that align digital interventions with corporate goals. Grounded in the KBV, this theme emphasizes that technology alone is insufficient. It must be embedded within a structured organizational architecture that governs how knowledge is captured, integrated and leveraged for value creation. These frameworks provide a roadmap for transitioning from ad hoc digital adoption to a mature, sustainability-oriented enterprise.

Emerging literature conceptualizes “digital intelligence” as a dynamic enabler that drives the integration of open innovation and circular economy strategies. By automating knowledge capture and facilitating cross-boundary collaboration, digital intelligence models help firms overcome barriers to external knowledge absorption (Huang and Zhou, 2025). For startups and small and medium-sized enterprises (SMEs), capability-based frameworks are essential for navigating resource constraints. Specifically, data governance and policy-driven innovation capabilities allow new ventures to align AI initiatives with long-term environmental goals (Santos et al., 2025). Green entrepreneurial orientation also acts as a strategic antecedent that enhances GKM practices and DC, which are moderated by BDA to yield sustainable competitive advantage (Cai et al., 2025). Furthermore, maturity models serve as diagnostic tools, classifying organizations along a sustainability continuum (Gupta et al., 2022; Vásquez et al., 2021). These models help leaders identify gaps in current KM infrastructure and prioritize investments in digital transformation, systematically directing knowledge assets toward environmental and economic targets (Shah, 2022; Sorout and Singh, 2025).

Responding to the paradigm shift toward Industry 5.0, this theme recenters the discourse around human well-being, creativity and social sustainability. It posits that sustainable KM is fundamentally a social process, where AI acts as a partner to augment human intelligence rather than replace it. The focus is on fostering HAI collaboration, promoting diversity, equity and inclusion, and ensuring that digital transformation leads to positive social outcomes and reduced technostress.

Empirically, the synergy between human creativity and AI capabilities strongly drives green collaborative innovation. However, this collaboration is most effective when guided by leadership with “green experience,” directing computational power toward environmentally meaningful solutions (Huang and Chen, 2025). The social dimension of sustainability is further reinforced by studies on AI adoption in sectors like journalism and healthcare, finding that KM factors such as technological affinity and trust are crucial for fostering sustainable use (Brescia et al., 2025; Li and Wang, 2025). In the context of social sustainability, KMS are being designed to democratize access to technology. For instance, low-code platforms empower nontechnical employees to contribute experiential knowledge to innovation processes, thereby fostering inclusivity and reducing the digital divide (Takahashi et al., 2024). Additionally, AI adoption in sustainability reporting is reshaping human and relational capital, creating new opportunities for engagement but also requiring careful management to maintain authenticity and stakeholder trust (Ciccola et al., 2025). Ultimately, this theme underscores that the resilient knowledge systems required for Industry 5.0 are built on a foundation of human-centric values (Lovrenčić, 2023).

This theme explores mechanisms through GKM and data analytics that translate into tangible ecological outcomes, such as eco-innovation and the circular economy, currently crucial emerging frontiers of the field. Rather than a mere environmental database, GKM acts as a dynamic catalyst converting raw data into green intellectual capital, comprising human, structural and relational capital, which is the primary engine for sustainable performance.

Empirically, GKM practices are the direct antecedents of green technological innovation. By systematically capturing and sharing environmental knowledge, organizations can enhance their green structural capital, which, in turn, mediates the relationship between GKM and sustainable performance (Khan et al., 2024; Zhang et al., 2025). BDA and AI capabilities moderate the impact of GKM on green innovation, suggesting that firms with superior data analytical skills are better able to leverage their green knowledge for radical innovation (Al Halbusi et al., 2025; Kumar et al., 2025). Furthermore, “green AI” is gaining traction, emphasizing energy-efficient algorithms (e.g. active learning and data set distillation) to minimize the carbon footprint of the knowledge discovery process (Salehi and Schmeink, 2024). From a resource orchestration perspective, aligning big data KM with competitive strategies allows firms to optimize resource allocation, ensuring digital investments directly support circular economy targets such as waste reduction and material recovery (Singh et al., 2024; Zhao and Niu, 2024). This logic also extends to social sustainability in education, where data-driven innovation facilitates sustainable learning models (Alismaiel, 2021).

The application of AI-driven KM for sustainability is highly context-dependent, with sectors leveraging distinct mechanisms to achieve their goals. This theme illustrates how universal KM principles are tailored to specific industrial environments. The literature demonstrates that successful implementation requires adapting technological interventions to the specific regulatory, operational and cultural realities of each sector.

In resource-intensive industries (e.g. oil, gas and forestry), the focus is on operational efficiency and proactive environmental management. For example, in the oil and gas sector of emerging economies, integrating AI and KM drives “interactive green innovation,” helping firms navigate complex geopolitical and environmental pressures (Abdulmuhsin et al., 2025). In Smart Forestry, AI-driven KM metamodels enhance decision-making regarding resource conservation and biodiversity, effectively managing the vast data generated by forest monitoring systems (Damaševičius and Maskeliūnas, 2025). Conversely, manufacturing emphasizes Industry 4.0 transformation, where the internet of things (IoT) and KMS integration drives operational excellence and product customization while minimizing waste (Karimi et al., 2025). In the service and SME sectors, the drivers are often related to policy and entrepreneurship. Government policy support is a crucial enabler strengthening the conversion of knowledge into sustainable entrepreneurship (Alshammakhi and Sheikh, 2025), while healthcare supply chains link AI maturity to achieving a sustainable competitive advantage through improved operational engagement (Aburayya, 2025). Finally, in the agricultural sector, KM models are transitioning from simple data processing to “wisdom-based” decision-making, crucial for adapting to climate change in arid regions (Abed et al., 2025).

To synthesize the diverse research streams identified above, Figure 8 illustrates the integrative conceptual framework derived from the thematic analysis. This framework maps the dynamic relationships among the three pillars, demonstrating how technological foundations are operationalized through organizational strategies to ultimately drive sustainable ecological outcomes.

The convergence of AI and KM catalyzes a paradigm shift toward sustainability-driven innovations. Grounded in the KBV, AI capability acts as a key driver for generating green intellectual capital, which, in turn, enhances DC and organizational creativity (Almheiri et al., 2025; Wang and Zhang, 2024). Concurrently, GKM frameworks enable systematic eco-innovation (Khan et al., 2024), while intelligent DSS and blockchain platforms empower transparent sustainability governance (Berndt et al., 2024; Singh et al., 2025). Furthermore, low-code platforms and GenAI democratize technology participation, fostering inclusive knowledge environments (Al-Emran et al., 2025; Takahashi et al., 2024).

However, a critical examination reveals significant limitations in how these opportunities are operationalized. Most studies assume ideal technological conditions, namely, mature infrastructures and high-quality data, which rarely exist in resource-constrained SMEs, startups and developing economies. The transition to human-centric models, such as Lean Service 5.0, is frequently impeded by sociocultural inhibitors, limited knowledge integration and a lack of structured communication flows (Bonamigo et al., 2025a). Furthermore, empirical evidence from traditional sectors reveals that cultural resistance and high competitiveness severely limit knowledge sharing and value co-creation across the supply chain (Bonamigo et al., 2025b). While studies frequently cite data silos, ethical considerations within the Industry 5.0 paradigm (e.g. AI bias, privacy protection and algorithmic transparency) are often reduced to superficial acknowledgments rather than rigorously examined mechanisms (Fantazy and Tipu, 2024; Brescia et al., 2025). The underlying assumption that technological solutions can seamlessly overcome organizational resistance reflects a techno-optimistic bias that ignores the complex, human-centric nature of sustainable knowledge systems.

Building on the critical assessment, fundamental knowledge gaps require immediate scholarly attention. Theoretically, the domain suffers from fragmentation. Studies typically draw on single lenses without establishing a clear theoretical hierarchy. This isolation prevents the development of comprehensive frameworks that can explain how AI-generated insights are converted into strategic knowledge assets through the KBV, operationalized in dynamic environments via DC and socially embedded through HAI collaboration from an STS perspective. Additionally, the treatment of multi-actor ecosystems often fails to examine the governance structures and power dynamics shaping collaborative AI–KM initiatives.

Methodologically and geographically, our findings align with the macro-level trends identified by Sorout and Singh (2025), confirming that the field’s reliance on purely quantitative approaches, cross-sectional surveys and the minimal representation from the Global South limit the capture of evolutionary dynamics and global generalizability. The field overlooks how power negotiations and unintended consequences emerge during AI implementation, which is a phenomenon that longitudinal tracking and advanced computational qualitative inquiry (e.g. NLP and text mining) could illuminate. Perhaps most critically, as AI systems gain influence over environmental decision-making, the field lacks a systematic investigation of how AI–KMS should be governed to ensure fairness, HAI collaborative synergy and accountability (Salehi and Schmeink, 2024).

To address these critical gaps, we propose six imperative research directions that challenge current assumptions while advancing both theoretical understanding and practical implementation. Table 2 provides an overview of each direction along with the underlying rationale. By embracing methodological diversity, theoretical integration, contextual sensitivity and ethical rigor, future research can move beyond techno-optimistic assumptions toward a more nuanced, inclusive and critically informed understanding of AI-driven KM for sustainability in the Industry 5.0 era.

This study systematically mapped the evolution of AI-driven KM for sustainable business through integrated bibliometric and thematic analyses. Bibliometrically, the domain demonstrates an accelerating velocity of knowledge creation driven by the strategic integration of AI and KM for sustainable business transformation, though this rapid expansion exposes critical vulnerabilities: a geographically siloed landscape that underrepresents the Global South, and a methodological overreliance on static cross-sectional surveys that obscure the dynamic organizational evolution of intelligent systems. Complementing these structural insights, the thematic analysis identified AI, KM and sustainability as anchor motor themes while highlighting green innovation and circular economy as critical emergent frontiers. By abstracting these thematic streams, we constructed a comprehensive tripartite framework encompassing technological foundations, organizational strategy and ecological outcomes. Ultimately, this research illuminates a fundamental paradigm shift from techno-centric applications toward human-centric ecosystems, providing a roadmap to leverage HAI synergies for inclusive and sustainable transformations.

This review makes several vital contributions to the theoretical development of this interdisciplinary domain. First, it addresses the prevailing theoretical fragmentation by establishing a clear theoretical hierarchy. By positioning the KBV as the primary analytical anchor, this study conceptualizes how AI-generated data are translated by KM architectures into green intellectual capital. Second, it integrates DC and STS as supplementary mechanisms to explain how these knowledge assets are dynamically reconfigured to sense environmental shifts (DC) and how they must be socially embedded through ethical governance (STS). Finally, this study significantly advances the discourse on Industry 5.0. By highlighting HAI collaboration, social inclusion and ethical governance as core themes, it theoretically solidifies the paradigm where technology serves to augment human creativity and social sustainability rather than simply automating processes.

To navigate the rapid evolution of AI-driven KM in sustainable business, our synthesis provides a practical framework focusing on technology implementation, cultural adaptation and ethical governance. For managers seeking to leverage digital transformation for sustainability, the identified maturity models and intelligent decision-support architectures serve as a valuable reference for technology selection and implementation. Crucially, this study underscores that successful AI–KM integration is not merely a technical challenge but a profound organizational transformation. Practitioners should prioritize overcoming sociocultural inhibitors, fostering cross-functional collaboration and nurturing digital intelligence alongside human experiential knowledge. For policymakers and industry regulators, the findings highlight the importance of establishing supportive governance. As intelligent platforms and blockchain-based ecosystems become central to supply chain traceability and environmental reporting, regulatory guidelines should encourage data privacy, algorithmic transparency and ethical AI deployment to ensure these technologies scale responsibly and inclusively, particularly for resource-constrained SMEs.

This review is subject to several inherent limitations that provide avenues for future inquiry. First, although the methodology included comprehensive searches across Scopus and WoS, this dual-database reliance may have omitted relevant studies published in other indices, gray literature or non-English languages. Second, despite updating the data retrieval to cover literature through the end of December 2025, the exponential pace of AI development dictates that emerging technological paradigms will continually reshape the landscape. Third, while the thematic content analysis was validated through team consensus, qualitative coding remains inherently interpretive and may have overlooked highly nuanced contextual factors. To overcome these boundaries, future research should embrace methodological diversity, expand geographic inclusivity and rigorously test the integrated frameworks proposed in this study across diverse empirical settings.

Furong Cai: Conceptualization, Methodology, Software, Formal analysis, Investigation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Ettore Bolisani: Conceptualization, Writing – review & editing, Supervision, Project administration. Maayan Nakash: Conceptualization, Methodology, Writing – review & editing. Tomas Cherkos Kassaneh: Methodology, Writing – review & Editing, Supervision.

The authors are grateful to the editor and the anonymous reviewers for their helpful comments and suggestions.