Generative Artificial Intelligence in Adaptive Social Manufacturing: A Pathway to Achieving Industry 5.0 Sustainability Goals

Abstract

As manufacturing transitions from Industry 4.0 to Industry 5.0, a critical challenge emerges in integrating Generative Artificial Intelligence (GAI) into adaptive social manufacturing to achieve sustainability goals. This transition reflects a paradigmatic shift from a technology-centric model focused on automation and efficiency toward a more holistic framework that embeds human-centricity and environmental responsibility into industrial systems. Whereas Industry 4.0 emphasizes digital innovation and productivity, Industry 5.0 seeks to align technological advancement with broader ecological and societal objectives. Despite advancements in automation and digitalization, existing frameworks lack a structured approach to leveraging GAI for environmental, social, and economic sustainability. This study explores the transformative role of GAI in adaptive social manufacturing, addressing the gap in the existing frameworks. Employing a multi-method research design, including content analysis, expert-driven validation, and system dynamics modeling, the study identifies nine key sustainability dimensions of Industry 5.0 and maps them to 17 GAI functions. The findings reveal that GAI significantly enhances adaptive social manufacturing by optimizing resource efficiency, promoting inclusivity, and supporting ethical governance. System dynamics analysis highlights the complex interdependencies between GAI-driven functions and sustainability outcomes, underscoring the need to balance technological innovation with human values. The research provides a novel framework for industries seeking to implement GAI in sustainable production systems, bridging theoretical insights with practical applications. Additionally, it offers actionable strategies to address challenges such as workforce adaptation, ethical AI governance, and adoption barriers, ultimately facilitating the transition toward Industry 5.0’s sustainability goals.

1. Introduction

Industry 5.0 has emerged as a transformative paradigm in recent years, offering a novel perspective on global industrial production. While Industry 4.0 was conceptualized in Germany with the advent of cyber-physical systems, Industry 5.0 introduces a more holistic view of the relationship between humans and machines, aiming to elevate this interaction toward greater synergy and co-creation. Although Industry 4.0 emphasized automation and cyber-physical systems, playing a pivotal role in the Fourth Industrial Revolution, Industry 5.0 moves beyond productivity and automation, integrating ethical, social, and environmental principles into production processes [1]. Since 2015, the concept has gained recognition in the academic literature as an innovative framework, with leading global companies striving to align their strategies with its human-centric, responsible, and sustainable principles [2,3]. Recent studies have further underscored the transformative role of GAI in operationalizing these Industry 5.0 principles across manufacturing systems. In this context, new developments have emerged in production systems, and one model that aligns with the core objectives of Industry 5.0 is social manufacturing. GAI plays a pivotal role in advancing Industry 5.0 by promoting sustainability, enabling adaptive manufacturing systems, and supporting human-centric production methods. Its integration into manufacturing processes fosters both efficiency and innovation while aligning with critical sustainability goals within the global economic landscape.

A key contribution of GAI to sustainability is its ability to optimize resource usage in manufacturing processes. For instance, GAI facilitates the design of adaptive manufacturing systems that enhance flexibility and responsiveness to market shifts and consumer demands [4,5]. This adaptability supports more sustainable practices by minimizing waste and enabling the creation of personalized products that better meet user preferences. Moreover, generative AI models simulate manufacturing scenarios to accurately estimate resource needs, thereby reducing unnecessary consumption and waste generation [6]. GAI also aids in maintenance, further improving resource efficiency by preventing equipment failures and minimizing downtime [7]. GAI also significantly contributes to human-centric production by enhancing collaboration between humans and machines through intelligent assistance systems that improve decision making. For example, detailed simulations generated by GAI allow workers to better visualize complex manufacturing tasks and meaningfully collaborate with AI-driven systems [8]. This synergy ensures that human intuition and creative problem solving are augmented—not replaced—by AI’s computational capabilities, thus fostering innovation and improving productivity within the sector [9]. With the shift toward a human-centric model, the goal becomes to create work environments in which AI enhances human strengths rather than rendering them obsolete [4,10].

In the broader context of the Global Value Chain, GAI plays a crucial role in enhancing productivity and significantly reducing trade costs, key to maintaining competitiveness in global markets [11]. By leveraging GAI, companies can improve supply chain alignment, ensuring that sustainability goals are met while optimizing cost-efficiency and responsiveness to dynamic market conditions. As such, GAI contributes to building resilient supply chains capable of adapting to sudden disruptions while adhering to environmental and social mandates [12]. In terms of environmental sustainability, GAI has proven to be a powerful tool in reducing the carbon emissions associated with manufacturing. By analyzing energy consumption patterns and optimizing production procedures, GAI helps firms significantly lower their carbon footprints [13,14]. Transitioning to greener practices is not merely a matter of compliance but a competitive necessity, especially as consumer expectations increasingly favor sustainability [15]. Thus, GAI’s role in driving ecological innovations that simultaneously enhance economic performance is central to achieving the objectives of Industry 5.0. Building on these advancements, the emergence of social manufacturing represents a natural extension of Industry 5.0 principles, particularly when integrated with GAI technologies that support decentralized, user-driven, and sustainable production models. This system enables users and producers to engage directly in the design and production processes, participate through social networks and digital platforms, and co-create value [4,5]. These systems facilitate customized production tailored to individual consumer needs while remaining aligned with the social and environmental aims of Industry 5.0. Although relatively new, less than a decade in existence, research indicates that such active participation enhances customer satisfaction and supports sustainability objectives through reduced waste [16]. Furthermore, incorporating cyber-physical systems, Internet of Things (IoT; sensor-equipped interconnected devices), and Artificial Intelligence (AI; systems mimicking human cognition) has allowed social manufacturing to swiftly adapt to environmental changes and evolving demands [17,18].

Concurrently, with the emergence of Industry 5.0 and social manufacturing, GAI was introduced in 2022 as a disruptive technology with wide-ranging applications across production systems, quickly capturing the attention of experts and industry stakeholders. Its capacity to autonomously generate creative content and simulate production processes fosters innovation and facilitates dynamic transformation within manufacturing systems [1]. For instance, in the customization of goods, GAI can generate models and product designs based on individual user preferences, meeting consumer needs while aligning with sustainability principles [19,20]. Moreover, GAI allows organizations to respond rapidly to market fluctuations by processing vast datasets and generating functional prototypes in significantly shorter timeframes. This leads to optimized resource usage, reduced environmental impact, and minimized waste production [21,22]. Despite these extensive capabilities, substantial challenges remain regarding the effective implementation of GAI in industrial settings. Many organizations face a lack of knowledge and insufficient skillsets to fully harness the technology. Additionally, due to inherent complexity, reliable frameworks and precise assessment tools for implementation have yet to be fully developed [4,23]. As a result, many industries remain hesitant, leading to slow adoption and integration. Ethical concerns also pose critical barriers, underscoring the necessity for robust governance frameworks to manage the societal and moral implications of GAI. These frameworks are essential to ensure transparency, social accountability, and trust among stakeholders [24], and they are fundamental for securing equitable and responsible outcomes in the application of GAI within social manufacturing.

This study addresses these challenges by employing a multi-method approach that integrates empirical assessment, expert-driven validation, statistical analysis, and system dynamics modeling. Unlike purely conceptual or systematic reviews, this research blends theoretical understanding with practical methodologies to offer a comprehensive, evidence-based perspective on GAI-enabled sustainability. Specifically, it employs system dynamics to construct a scalable and scientifically grounded framework for implementing GAI in social manufacturing, advancing the broader sustainability objectives of Industry 5.0. This study emphasizes the role of GAI in reducing waste, optimizing resources, and enhancing human–machine collaboration. Furthermore, it identifies the knowledge gaps hindering GAI adoption and proposes structured solutions to address these barriers. The proposed framework evaluates essential factors such as human–machine interaction, AI flexibility, and the automation capabilities required for effective integration. Ultimately, this study aims to demonstrate how these factors contribute to the successful and sustainable adoption of Industry 5.0.

2. Literature Review

The transition from Industry 4.0 to Industry 5.0 marks a fundamental shift in industrial systems. Building on the digital foundations of Industry 4.0, this new paradigm expands its focus to include broader societal goals, emphasizing the synergy between human capabilities and technological advancements to achieve sustainable and innovative outcomes [25]. This shift addresses the limitations of previous industrial eras that prioritized technological efficiency over human and environmental considerations [26,27]. Effective adoption of Industry 5.0 requires a workforce that balances technical abilities with adaptability and strategic thinking. It places high importance on human adaptability and the ability to collaborate effectively with intelligent systems to foster productivity and sustainability [25]. Additionally, integrating cyber-physical systems, IoT, and big data is essential for creating interconnected, efficient systems responsive to real-time demands [28,29]. The adoption of sustainable practices within Industry 5.0 is critical as industries face mounting pressures to reduce their environmental impact. Ref. [30] argues that circular economy principles fostered during Industry 4.0 provide a robust framework for transitioning toward sustainable business models in Industry 5.0. Such practices are fundamental not only as a means of ecological responsibility but also as a pathway to harmonize industrial growth with environmental objectives [27]. However, this transition also necessitates infrastructural advances to support the seamless integration of digital and cyber-physical technologies. Strategic planning must therefore be tailored to each organization’s goals and capabilities, as Industry 5.0 is not a one-size-fits-all approach. Companies must assess their core competencies and develop customized strategies for an effective transition [31].

Emerging technologies, particularly GAI, play a vital role in realizing Industry 5.0 goals by enhancing productivity, fostering innovation, and supporting sustainable practices across diverse sectors. Convergence technologies, as discussed by He and [23], enhance industrial resilience by dismantling information barriers and facilitating coordinated development across industries, including newer sectors like information technology. Furthermore, technological advancements in logistics, such as those highlighted by [32], streamline inventory and warehouse operations [33,34,35,36], contributing to sustainability by reducing waste and optimizing resource utilization. Technological advancements in logistics, such as the use of generative AI for agile production decisions, operational resilience, and enhanced data consistency, have revolutionized manufacturing responsiveness and efficiency. For instance, generative AI facilitates real-time data analysis to improve decision making on the shop floor, enables the rapid identification of risks to enhance resilience, and ensures high-quality, consistent data to support predictive logistics and streamlined operations. These advancements support Industry 5.0’s vision of a resilient, sustainable, and interconnected industrial ecosystem that aligns with environmental stewardship. GAI, with its capacity to produce content and systems that mimic human creativity, is a critical technology in the transition to Industry 5.0 [37]. It has revolutionized various sectors by expediting product development, enabling rapid prototyping, and shortening time-to-market, helping companies maintain agility and innovation [23,38]. GAI facilitates the creation of synthetic data that closely replicates real-world scenarios, minimizing risks in early development stages without requiring extensive data collection [37]. Additionally, manufacturers can optimize resource use and reduce waste by employing generative design techniques, promoting economic and environmental sustainability in alignment with Industry 5.0 objectives. Beyond operational efficiency, GAI reshapes workforce dynamics by automating routine tasks and providing intelligent assistance [38], allowing employees to focus on higher-order decision making and creative problem solving [39]. Moreover, GAI supports workforce development by facilitating personalized learning experiences, equipping employees with relevant skills for modern manufacturing environments, and promoting adaptability in an evolving industry [40]. Such advancements address the labor demands of Industry 5.0 and underscore the importance of fostering a knowledgeable, adaptable workforce capable of thriving in collaborative, technology-driven settings [41].

In this context, qualitative analysis has emerged as a robust research methodology to explore the sustainability implications of emerging technologies, including AI and Industry 5.0. Qualitative analysis facilitates the systematic identification of themes, patterns, and categories in textual data, allowing for a deeper understanding of how sustainability is operationalized across technological and industrial systems. For example, ref. [42] applied qualitative analysis to educational content and highlighted the under-representation of sustainability concepts, particularly environmental dimensions. In the domain of AI, QCA has been used to assess the off-site manufacturing literature [43], examine sustainability indicators in carbon-intensive sectors [39,44], and analyze corporate sustainability reports through thematic categorization [45]. This methodology has also provided valuable insights into sustainability reporting practices across global industries, revealing how organizations align their strategies with Sustainable Development Goals (SDGs) [46,47].

Specifically, in AI-integrated industries, qualitative analysis enables scholars to investigate how technologies such as GAI support sustainability objectives by uncovering latent themes such as stakeholder engagement, carbon footprint reduction, and ethical compliance [12,48]. In this regard, qualitative approaches complement quantitative metrics by offering context-rich insights into technological adoption and its broader implications. The flexibility of qualitative analysis also makes it well suited to analyzing the evolving dynamics of Industry 5.0, which necessitates an understanding of social, human-centric, and environmental priorities [49].

In the context of social manufacturing, which emphasizes decentralized production through social networks, system dynamics offers a valuable framework for modeling complex interactions and enabling real-time decision making. Unlike traditional methodologies, which often fall short in capturing the evolving interactions within social manufacturing networks, system dynamics allows for a comprehensive analysis of collaborative efficiency and resource allocation [50]. This approach facilitates the optimization of production processes and supports the development of social value chains that enhance value creation for all participants [51]. The interconnected nature of social manufacturing demands a flexible and responsive production environment. By incorporating cyber-physical systems and social sensors, manufacturers can monitor and adjust tasks in real time, ensuring that production aligns with shifting consumer demands and market conditions [52]. System dynamics also aid the digital transformation of traditional manufacturing sectors by enabling simulations that reveal optimal configurations for integrated production systems. Research by [53] shows how system dynamics models help manufacturers understand the impact of digital integration on productivity and sustainability, supporting the adaptive, interconnected, and collaborative goals of Industry 5.0. However, the application of GAI and system dynamics in manufacturing is not without challenges, especially regarding ethical and privacy concerns. Ensuring transparency and accountability in GAI systems is crucial to gaining stakeholder trust and facilitating widespread adoption [54]. The moral implications of GAI are particularly significant in areas such as content creation, intellectual property, and consumer engagement [55]. As technologies like Generative Adversarial Networks gain traction, questions regarding the authenticity and ownership of AI-generated content become increasingly relevant [56].

In sectors like marketing and fashion, GAI enables highly personalized consumer experiences that enhance customer loyalty and reshape consumer expectations [57,58]. These applications highlight the transformative potential of GAI, but also underscore the need for robust ethical frameworks. Therefore, a qualitative understanding of these dynamics, as supported by qualitative analysis, is vital in ensuring the responsible adoption of emerging technologies. The literature review reveals a notable gap in the integrated application of GAI and system dynamics within social manufacturing environments. While both have independently enhanced manufacturing processes, limited research explores their combined role in improving adaptability and enabling real-time decision making. Specifically, the existing literature lacks comprehensive frameworks in which GAI supports adaptive capacities in social manufacturing through modeling. Addressing this gap, the present study demonstrates how integrating GAI with system dynamics can establish continuous feedback loops and support data-driven decision making, thus fostering more resilient and responsive manufacturing systems aligned with Industry 5.0’s sustainability goals. The insights synthesized here form the foundation for the study’s empirical design, informing expert validation, statistical analysis, and modeling, thereby enabling a structured evaluation of GAI’s contribution to sustainable industrial transformation.

3. Theoretical Framework

System dynamics serves as a key analytical framework for studying the complex interactions within manufacturing systems, particularly in adaptive social manufacturing. This approach enables researchers to capture intricate relationships among system components and evaluate the long-term impacts of various decisions [59]. Central to system dynamics is the concept of examining work-in-process levels, order quantities, and production rates, which can be modeled through differential or difference equations to reflect the continuous nature of manufacturing systems. By accommodating stochastic fluctuations in production capacity and demand, system dynamics provides a strong foundation for understanding and optimizing production processes, especially in adaptive systems where dynamic changes are prevalent [60,61].

In the evolving framework of Industry 5.0, system dynamics modeling plays a crucial role in enhancing sustainable manufacturing and fostering social manufacturing approaches. As industries shift towards a model that emphasizes collaboration between human and machine capabilities, adopting system dynamics provides significant advantages in understanding complex behaviors over time and predicting the implications of different manufacturing strategies. Various studies illustrate the impact of system dynamics on manufacturing productivity and sustainability. For instance, ref. [62] outlines the application of system dynamics in improving productivity through Causal Loop Diagrams (CLDs), which help visualize interdependencies within manufacturing systems and identify leverage points for enhancing efficiency. Similarly, the research presented by [63] highlights how manufacturing flexibility and multi-criteria optimization models can enhance sustainability within manufacturing systems.

The complexities in manufacturing systems can be broadly classified into static and dynamic categories. Static complexity arises from the physical structure of the system, while dynamic complexity emerges from unpredictable behaviors over time, especially in regulated environments [64]. This distinction is essential for developing simulation frameworks that capture the multifaceted nature of manufacturing operations. For instance, modular hybrid simulation frameworks have been proposed to address both static and dynamic complexities, enhancing the design and analysis of manufacturing systems. In adaptive social manufacturing, system dynamics allows for the coordinated management of disturbances such as order changes, processing delays, and equipment malfunctions, thereby maintaining the system stability and adaptability [14,61].

Moreover, specific applications of system dynamics have demonstrated significant results in sectors like petrochemicals. Ref. [65] utilizes a system dynamics approach to develop scenarios balancing emission reduction and profitability, emphasizing its utility in crafting sustainable pathways for industry. In the eco-industry context, ref. [66] examines long-term environmental and economic effects using system dynamics modeling, stressing the importance of dynamic modeling tools in managing the footprint of industrial activities.

Additionally, system dynamics extends to reliability modeling, which has evolved from static to dynamic and multistate frameworks. Traditional reliability models, often limited in scope, cannot fully capture the complexities of modern manufacturing systems, which require advanced frameworks capable of accounting for multiple operational states and uncertainties [67]. This capability is particularly relevant in adaptive social manufacturing, where flexibility and resilience to environmental changes are vital. The systematic approach of system dynamics supports researchers in modeling complex system interactions and continuously evaluating adjustments and decisions to optimize long-term outcomes [68].

Social manufacturing, a core component of Industry 5.0, also benefits from system dynamics. Sari et al. illustrate an integrated production system that emphasizes collaboration among distributed manufacturing resources, which enhances production efficiency and sustainability [69]. Furthermore, research by [70] investigates the integration of social factors in manufacturing systems, applying system dynamics to examine social sustainability metrics alongside traditional sustainability measures. The intersection between system dynamics and social manufacturing enables manufacturers to engage with consumers and communities throughout the production life cycle. In this context, ref. [71] explores collaborative agglomeration in the tourism industry using system dynamics to understand how industrial and cultural synergies can promote sustainable practices. Such integrative approaches are essential in navigating the complexities inherent in modern manufacturing landscapes, where stakeholder interactions and resource sharing define successful performance.

Finally, integrating dynamic analysis and simulation tools allows researchers to simulate complex interactions within adaptive systems. When combined with autonomous planning methodologies, these analyses enable the design of manufacturing systems capable of rapid and flexible responses to environmental shifts. Specifically, source [72] introduces simulation-based optimization techniques that support dynamic decision making under uncertainty, ref. [73] presents a multi-agent planning approach tailored for manufacturing reconfiguration, and ref. [74] demonstrates the role of GAI-enhanced feedback loops in adapting production systems to real-time environmental data. Collectively, these studies underscore that system dynamics modeling is not merely a theoretical construct but a practical toolkit for industries striving to adapt in the context of Industry 5.0. By facilitating a comprehensive understanding of interconnected manufacturing processes, this methodology fosters an ecosystem where sustainability and social responsibility are integrated into the fabric of operations.

4. Method

This study adopted a rigorous, multi-phase research methodology to explore the integration of GAI in enhancing sustainability within adaptive social manufacturing systems, an emerging paradigm under the Industry 5.0 framework. The methodological design followed a sequential structure, with each phase building upon the preceding one to ensure conceptual continuity, empirical rigor, and comprehensive alignment with the research objectives. The initial phase involved an in-depth qualitative content analysis, conducted using Leximancer 5.0 software, which enabled concept mapping and co-occurrence analysis across a curated corpus of academic literature. Articles were selected based on a predefined set of inclusion criteria, including peer-reviewed status, relevance to Industry 5.0 and sustainability, English language, and publication between 2018 and 2025. Search terms used included combinations of “Industry 5.0”, “sustainability”, “AI in manufacturing”, and “human-centric production.” Studies lacking explicit reference to sustainability dimensions or AI applications were excluded. In addition to Leximancer-driven mapping, the data underwent open and axial coding to refine thematic categories and ensure conceptual clarity. These codes were validated through triangulation with established literature and expert feedback, resulting in the identification of nine core sustainability dimensions: environmental resilience, energy optimization, circular economy practices, social inclusion, human–machine collaboration, digital ethics, policy adaptability, economic competitiveness, and innovation capability.

To empirically validate the relevance and completeness of these dimensions, the second phase engaged 15 seasoned industry experts, each with at least ten years of professional experience across fields such as advanced manufacturing, sustainability consulting, smart logistics, and industrial AI implementation. Experts were selected for their disciplinary and geographic diversity, with professional affiliations spanning Europe, North America, and Asia, although all participants were currently based in Australia. A three-round, Delphi-based virtual consultation was conducted, with each round lasting approximately one hour. Experts critically assessed each sustainability dimension based on criteria such as clarity, practical relevance, and strategic alignment with Industry 5.0 objectives. Their feedback was collected through digital collaboration platforms, analyzed using thematic content analysis, and synthesized to refine and finalize the framework. To ensure analytical rigor, the process incorporated both manual coding and inter-rater reliability checks using Cohen’s Kappa, where values above 0.70 across all validation rounds indicated substantial agreement and methodological robustness.

In the third phase, the study shifted its focus to identifying and validating GAI functions that could influence these sustainability dimensions. A large-scale Focus Group Discussion (FGD) approach was employed, engaging 130 domain experts in applied AI, digital transformation, and industrial engineering. Participants were recruited through a specialized Australian-based professional network dedicated to AI applications in sustainable industry. The FGD process was structured across three intensive sessions, each lasting three hours. In the first session, participants were divided into five subgroups of 26, each led by an academic moderator. The initial 60 min involved structured brainstorming on Industry 5.0 challenges and the role of AI in addressing these challenges. Experts collaboratively mapped their insights on digital whiteboards. The subsequent 90 min focused on open and axial coding, where participants identified, grouped, and categorized candidate AI functions based on their operational relevance and potential impact.

The second session involved a critical review of the preliminary AI function list. Subgroups reconvened to reassess each function in terms of relevance, redundancy, and clarity. Functions deemed duplicative or marginally impactful were merged or excluded. Participants then conducted a quantitative evaluation using a five-point Likert scale to rate each AI function’s impact on the nine sustainability dimensions. In the third session, the Likert ratings were aggregated and analyzed. Experts engaged in structured deliberation to prioritize AI functions based on collective consensus and aggregated scores. Through an iterative refinement process, the final list of 17 GAI functions was established based on their strategic significance, applicability, and perceived sustainability impact. While the participants represented a wide array of national backgrounds, all were based in Australia, which helped ensure that the identified AI functions addressed region-specific sustainability challenges while leveraging globally informed expertise.

In the fourth phase, the research employed advanced statistical analysis to evaluate the relationship between the 17 AI functions and the validated sustainability dimensions. Each expert rated the impact of all AI functions across the nine dimensions using a standardized five-point Likert scale (1 = very low impact, 5 = very high impact). The collected data underwent Multivariate Analysis of Variance (MANOVA) to simultaneously assess the effects of multiple independent variables (AI functions) on several dependent variables (sustainability dimensions). MANOVA was chosen for its capability of detecting complex multivariate relationships and interaction effects. To accommodate potential non-normality in the ordinal data, the Friedman test, a robust non-parametric alternative, was also employed. This dual-analysis strategy enhanced the reliability of statistical inferences and enabled the prioritization of GAI functions based on statistical significance and aggregate sustainability impact. The final phase aimed to explore the systemic interconnections between GAI functions and sustainability outcomes using system dynamics modeling. A CLD was developed through a three-stage expert validation process using Vensim PLE software version 10.2.2. From the initial 130 experts, 48 were selected for this phase based on five predetermined criteria: (1) at least five years of professional experience in industrial AI or sustainability, (2) active involvement in at least one critical industrial sector (e.g., smart manufacturing, energy systems, logistics), (3) demonstrated expertise in system dynamics modeling, (4) interdisciplinary background across technology, policy, and sustainability, and (5) prior contributions to scholarly or applied research in relevant domains.

In the first CLD session, the selected experts were divided into four subgroups. Each subgroup collaboratively mapped causal links between the 17 AI functions and sustainability dimensions using digital whiteboards. The result was a preliminary CLD with identified reinforcing (R) and balancing (B) loops. In the second session, the initial model underwent refinement: experts revisited the structure, removed redundant links, and proposed additional interdependencies. A structured Likert voting mechanism (scale of 1–5) was implemented to rate the strength of each causal link. Only links rated ≥4.0 were retained to ensure consensus-based rigor. In the third session, the refined CLD was presented to all 48 experts, who then identified key leverage points where specific AI functions exhibited systemic influence. To verify the reliability of expert evaluations, Cohen’s Kappa (κ) was calculated to assess inter-rater agreement for each identified causal relationship. The resulting κ values ranged from 0.72 to 0.81, indicating substantial to near-perfect agreement, thereby confirming the reliability and robustness of the model. Figure 1 illustrates the sequential research design and methodological integration across all six phases, highlighting the empirical and analytical procedures used to generate, validate, and synthesize the study’s findings.

5. Results

5.1. Content Analysis of Industry 5.0 Sustainability Goals

The purpose of this content analysis was to identify the sustainability dimensions that uniquely characterize Industry 5.0, setting it apart from the efficiency-focused paradigm of Industry 4.0. Through a rigorous thematic analysis of selected scholarly articles, this study aimed to reveal the specific sustainability goals that align with Industry 5.0’s human-centric, ethical, and ecological principles. Utilizing Leximancer for concept mapping, this analysis offers a structured and objective view of the sustainability themes essential to Industry 5.0’s framework. The content analysis began with a systematic literature review conducted across major academic databases, including Scopus and Web of Science, with a focus on peer-reviewed publications from the past seven years. A total of 104 articles were selected based on search terms explicitly related to Industry 5.0 and sustainability, such as “Industry 5.0 sustainability goals”, “human-centric manufacturing”, and “ethical AI in manufacturing”. Each article was reviewed for relevance to Industry 5.0’s distinct sustainability framework.

The analysis identified nine primary themes, each representing a distinct aspect of sustainability within Industry 5.0. Environmental sustainability, addressed in 25 articles, emphasizes carbon emission reduction, resource recycling, reuse, and supply chain management, underscoring Industry 5.0’s commitment to minimizing environmental impact. Social sustainability, supported by 17 articles, focuses on enhancing employee welfare, promoting social responsibility, and supporting community development. Economic sustainability, noted in 10 articles, emphasizes reducing costs, increasing productivity, and fostering flexible business models. Ethical sustainability, examined in 10 articles, addresses equitable resource distribution and transparency in decision making. Technological sustainability, referenced in eight articles, highlights the role of advanced technologies, such as the Internet of Things and big data, in enhancing efficiency and reducing environmental impact. Cultural sustainability, covered in eight articles, emphasizes preserving cultural identity and local values within production processes. Supply chain sustainability, discussed in eight articles, focuses on transparency within supply chains and managing product life cycles. Human sustainability, examined in eight articles, pertains to improving employee health and safety and creating secure work environments. Lastly, managerial sustainability emerges in 10 articles, highlighting the development of sustainable management strategies and international cooperation across supply chains. To examine the interactions and relationships among these themes, the Leximancer software (LexiDesktop5) was employed, producing a concept map based on term frequency and co-occurrence analysis. The resulting map, shown in Figure 2, visually illustrates the primary clusters identified, such as Social Welfare and Inclusivity, Ethical AI Integration, and Environmental Efficiency. The map also highlights a strong link between environmental sustainability and technological sustainability, underscoring the importance of green and intelligent technologies in Industry 5.0 for minimizing environmental impact.

To provide a concise overview of these dimensions, the following table summarizes their definitions, key characteristics, and the number of academic sources supporting each theme. This summary serves as a foundation for understanding how these dimensions collectively contribute to achieving the sustainability goals of Industry 5.0 (

Table 1. Sustainability dimensions in Industry 5.0 *.

	Sustainability Dimension	Definition and Characteristics	Sources
1	Environmental sustainability	Focuses on reducing carbon emissions, resource recycling, and supply chain management to minimize environmental impacts.	25 articles
2	Social sustainability	Emphasizes enhancing employee welfare, fostering social responsibility, and supporting community development.	17 articles
3	Economic sustainability	Involves cost reduction, productivity enhancement, and the development of flexible business models.	10 articles
4	Ethical sustainability	Addresses equitable resource distribution and transparency in decision-making processes.	10 articles
5	Technological sustainability	Highlights the role of advanced technologies, such as the IoT and big data, in improving efficiency and reducing environmental impact.	8 articles
6	Cultural sustainability	Aims to preserve cultural identity and local values within production processes.	8 articles
7	Supply chain sustainability	Focuses on ensuring transparency in supply chains and managing product life cycles effectively.	8 articles
8	Human sustainability	Relates to improving employee health and safety and fostering secure and healthy work environments.	8 articles
9	Managerial sustainability	Highlights the development of sustainable management strategies and international cooperation across supply chains.	10 articles

* Details are provided in Appendix A.

).

Following the thematic analysis and concept mapping using Leximancer, a validation phase was conducted to ensure that the sustainability dimensions were clear, practically relevant, and applicable to adaptive social manufacturing. This phase involved 15 industry experts specializing in advanced manufacturing and sustainability consultancy, all of whom had extensive experience in Industry 5.0-related sustainability practices. The validation process consisted of three structured sessions, each lasting one hour, where experts critically assessed the sustainability dimensions identified through content analysis. In the first session, experts were provided with detailed descriptions of each sustainability dimension, including definitions and conceptual justifications derived from the thematic analysis. They were asked to evaluate the clarity, industry relevance, and practical applicability of each dimension, providing written feedback that was later coded into thematic categories. In the second session, expert feedback from the first session was analyzed using open and axial coding, identifying recurring themes and concerns related to overlapping, ambiguous, or redundant dimensions. Experts were then engaged in a structured discussion, where they collectively assessed dimensions that needed refinement, merging, or redefinition. Through this collaborative process, some sustainability dimensions were consolidated to remove redundancies, while others were expanded or reworded to improve conceptual clarity. In the third and final session, experts reviewed the revised sustainability dimensions and provided quantitative validation through a five-point Likert scale, rating the importance and clarity of each dimension. To ensure inter-expert agreement, a Cohen’s Kappa test was conducted, yielding a high consensus score (Kappa = 0.80), indicating strong alignment among experts. The final validated framework consisted of nine sustainability dimensions, which had been refined through expert-driven evaluation and systematic feedback analysis. This validation process ensured that the finalized sustainability dimensions were both theoretically sound and practically applicable, providing a robust foundation for assessing sustainability in Industry 5.0. The validated framework reflects expert consensus, making it a reliable reference for sustainable industrial practices within adaptive social manufacturing (

Table 2. Validation of sustainability dimensions in Industry 5.0.

	Sustainability Dimension	Expert Feedback	Action Taken	Final Validated Sustainability Dimension	Agreement Score (Mean ± SD/Cohen’s Kappa)
1	Environmental sustainability	Clear and highly relevant	No change	Environmental sustainability (EnS)	4.7 ± 0.5/0.82
2	Social sustainability	Clear and highly relevant	No change	Social sustainability (SoS)	4.5 ± 0.6/0.79
3	Economic sustainability	Clear and highly relevant	No change	Economic sustainability (EcS)	4.6 ± 0.4/0.81
4	Ethical sustainability	Clear and highly relevant	No change	Ethical sustainability (EthS)	4.6 ± 0.5/0.80
5	Technological sustainability	Clear and highly relevant	No change	Technological sustainability (TS)	4.8 ± 0.3/0.85
6	Cultural sustainability	Clear and highly relevant	No change	Cultural sustainability (CS)	4.3 ± 0.6/0.77
7	Supply chain sustainability	Clear and highly relevant	No change	Supply chain sustainability (SCS)	4.5 ± 0.5/0.79
8	Human sustainability	Clear and highly relevant	No change	Human sustainability (HS)	4.5 ± 0.6/0.79
9	Environmental sustainability	Clear and highly relevant	No change	Managerial sustainability (MS)	4.4 ± 0.5/0.78

).

5.2. Expert Engagement and Validation

This study employed a FGD approach with 130 industry experts across three structured three-hour sessions to identify and validate 17 key GAI functions for Industry 5.0 sustainability. In the first session, experts were divided into five subgroups, where they discussed sustainability challenges and the role of AI using digital whiteboards. This was followed by open coding, where AI functions were categorized based on similarities. The second session focused on reviewing and refining the preliminary list, eliminating redundancies, and conducting a five-point Likert scale assessment to evaluate each function’s impact on nine sustainability dimensions. In the final session, aggregated ratings informed the prioritization of 20 AI functions, from which the 17 most impactful and strategically relevant functions were selected through an iterative consensus-building process. To ensure the reliability of expert judgments and inter-group agreement, a Cohen’s Kappa test was conducted, confirming a high level of consensus among expert groups. Additionally, MANOVA was applied to quantify the statistical significance of expert ratings, ensuring that the final selection was not only consensus-driven but also statistically validated. The observed agreement percentages ranged between 85% and 92%, while the Kappa coefficient (κ) values fell between 0.625 and 0.733, indicating substantial inter-rater reliability across all groups. Given that κ values above 0.61 denote substantial agreement, these results confirm the robustness of expert evaluations. The standard error remained below 0.045 in all cases, and 95% confidence intervals further validated the reliability of these findings (

Table 3. Inter-rater reliability analysis using Cohen’s Kappa statistics.

Groups	Raters	Rated Items	Observed Agreement (%)	Expected Agreement (%)	Kappa Coefficient (κ)	SE	95% Confidence Interval (CI)	Interpretation
1	26	50	85	60	0.625	0.045	0.537–0.713	Substantial agreement
2	26	50	90	65	0.714	0.042	0.632–0.796	Substantial agreement
3	26	50	92	70	0.733	0.040	0.654–0.812	Substantial agreement
4	26	50	88	62	0.684	0.043	0.600–0.768	Substantial agreement
5	26	50	87	61	0.667	0.044	0.581–0.753	Substantial agreement

).

The inter-rater agreement results obtained in this study, specifically the Cohen’s Kappa values ranging from 0.625 to 0.733, indicate substantial agreement among expert panels. These values are consistent with findings in the recent literature evaluating expert consensus on AI applications in sustainability-focused domains. For instance, ref. [75] reported similar Kappa coefficients in assessing GAI’s role in the construction industry using multi-criteria decision-making techniques, reinforcing the validity of structured expert involvement. Additionally, ref. [76] emphasized the importance of expert evaluations in the sustainable energy sector, where GAI-driven resource optimization required cross-disciplinary validation. These parallels confirm that the levels of agreement in our study align with established practices in AI-related sustainability assessments. Moreover, the observed consistency reflects the robustness of our methodological approach, including structured FGDs and multi-stage validation processes, similar to those advocated in broader Industry 5.0 research [77,78,79]. This comparison with the existing literature substantiates the reliability of the findings presented in Table 3 and reinforces their generalizability across industrial contexts.

By integrating a multi-stage expert discussion approach, coupled with qualitative thematic coding and quantitative statistical validation, this study ensured that the final 17 AI functions were identified based on scientific rigor, industry consensus, and empirical validation. The expert panel consisted of 130 specialists with 2–10 years of experience in applied AI for industrial sustainability, ensuring a comprehensive and balanced perspective. These experts, actively engaged in GAI-driven solutions for real-world manufacturing environments, provided critical insights into the intersection of GAI, circular economy principles, and adaptive production systems. Their inclusion ensured that the study captured not only the technical potential of AI but also its feasibility and impact on sustainable industrial ecosystems. The recruitment strategy, based on an Australian-based professional forum, ensured that regionally relevant sustainability challenges were incorporated while still benefiting from the global expertise and interdisciplinary knowledge of participants.

Seventeen GAI Applications through Expert Interviews:

Predictive Maintenance (PM): This uses real-time data from sensors and IoT devices to monitor the condition of equipment and predict potential failures. By analyzing patterns in equipment behavior, it identifies anomalies and forecasts maintenance needs, reducing unplanned downtime and preventing catastrophic failures. PM extends the equipment lifespan and reduces material waste, aligning with environmental sustainability. It also minimizes costs, contributes to economic sustainability, and supports managerial sustainability by improving operational planning.

Resource Optimization Algorithms (ROAs): These algorithms analyze production data to identify inefficiencies in the use of resources such as energy, raw materials, and water. They recommend optimal allocation strategies to maximize resource efficiency while minimizing waste. ROAs reduce energy consumption and material waste, contributing to environmental sustainability. They enhance cost-efficiency, support economic sustainability, and improve decision-making processes for managerial sustainability.

Sustainable Materials Development (SMD): GAI accelerates the discovery of sustainable materials by simulating chemical properties and predicting the performance of new compounds. It helps industries identify and adopt eco-friendly alternatives to traditional materials. SMD advances technological sustainability by fostering innovation in material science and supports environmental sustainability by reducing reliance on non-renewable resources.

Workforce Engagement and Training Platforms (WETPs): GAI-powered platforms provide personalized training modules for employees, identifying skill gaps and offering tailored learning experiences. These systems use gamification and real-time feedback to enhance learning outcomes. WETPs improve workforce adaptability and inclusivity, aligning with human and social sustainability, enhance job satisfaction, and ensure employees are equipped to handle Industry 5.0 challenges.

Community Engagement Tools (CETs): These leverage GAI to foster collaboration between industries and local communities. These tools collect community feedback, assess societal needs, and enable participatory design processes. CETs promote cultural sustainability by preserving local values and traditions. They strengthen social sustainability by involving communities in decision making and ensuring ethical practices.

Supply Chain Optimization (SCO): SCO tools analyze data across the supply chain to streamline logistics, manage inventory, and enhance transparency. These systems enable real-time monitoring and predictive adjustments to prevent disruptions. SCO reduces waste and inefficiencies, supporting economic and environmental sustainability. It enhances trust and ethical practices in supply chains, contributing to supply chain and ethical sustainability.

Innovative Business Models (IBMs): GAI facilitates the design of IBMs by analyzing market trends, consumer behaviors, and economic patterns. It supports the adoption of circular economy principles, such as reuse and recycling. IBMs advance economic sustainability through new revenue streams and operational efficiencies. They encourage environmental sustainability by promoting resource reuse and waste reduction.

Transparent Decision Frameworks (TDFs): TDF tools use GAI to provide clear and auditable decision-making processes. By analyzing large datasets, they offer justifications for business choices, increasing transparency and accountability. TDFs enhance ethical sustainability by fostering trust among stakeholders. They support managerial sustainability by improving governance and decision making.

Bias Monitoring Tools (BMTs): BMTs detect and mitigate biases in algorithms, processes, and data inputs. These tools ensure fairness in GAI applications by identifying and correcting discriminatory patterns. BMTs support ethical sustainability by ensuring inclusivity and fairness. They promotes diversity in decision making and resource allocation.

Cyber-Physical System Integration (CPSI): CPSI integrates physical manufacturing processes with cyber systems, enabling real-time monitoring, automation, and coordination. These systems create a digital twin of manufacturing environments for optimized control. CPSI advances technological sustainability by enabling the seamless integration of AI and IoT. It enhances supply chain sustainability by improving transparency and adaptability.

Automatic Troubleshooting (AT): AT systems use GAI to diagnose and resolve technical issues without human intervention. By identifying root causes and suggesting solutions, they reduce operational downtime. AT supports human sustainability by reducing stress on workers. It enhances technological sustainability through automation and improves managerial sustainability by ensuring consistent operations.

Customized Product Designs (CPDs): GAI enables the creation of personalized product designs based on individual preferences, cultural values, and functional requirements. This application tailors production to specific consumer needs. CPDs preserve cultural identity and promotes cultural sustainability. They enhance consumer satisfaction, contributing to economic sustainability.

Dynamic Supply Chain Modeling (DSCM): DSCM uses GAI to adapt supply chain strategies to changing market conditions, such as demand fluctuations, geopolitical events, or climate impacts. It ensures resilience and efficiency in supply networks. DSCM optimizes supply chain operations, supporting supply chain, environmental, and managerial sustainability, reduces waste, and enhances responsiveness.

Safety Simulation Systems (SSSs): SSSs use AI-driven simulations to predict and mitigate workplace safety risks. These systems model potential hazards and suggest preventive measures to ensure employee well-being. SSSs improve workplace conditions, aligning with human sustainability. They reduce accident-related costs, indirectly contributing to economic sustainability.

Ergonomic Solutions (ESs): ESs leverage GAI to design workplaces and tools that prioritize human comfort and efficiency. These solutions analyze worker movements to minimize strain and injuries. ESs enhance human sustainability by promoting health and safety, and support social sustainability by fostering inclusive work environments.

AI-Enhanced Decision Support Systems (AIDSSs): AIDSSs provide managers with real-time data insights and scenario analyses, aiding in strategic planning and decision making. These systems reduce uncertainty and improve business agility. AIDSSs support managerial sustainability through data-driven governance. They enhance economic sustainability by optimizing resource allocation and reducing risks.

Scenario Planning Tools (SPTs): SPTs simulate various future scenarios, enabling businesses to anticipate and prepare for uncertainties such as market changes, technological disruptions, or environmental challenges. SPTs strengthen managerial sustainability by improving long-term resilience and adaptability. They help organizations align strategies with sustainability goals.

5.3. Evaluation Framework and Statistical Analysis

The descriptive statistics in

Table 4. Descriptive statistics of AI functions.

GAI Applications	Mean	SD	Skewness	Kurtosis
PM	3.41	1.139	−0.058	−1.042
ROA	3.42	1.289	−0.482	−0.814
SMD	3.12	1.314	0.004	−1.077
WETP	3	1.479	−0.002	−1.44
CET	3.2	1.271	−0.124	−0.901
SCO	3.04	1.347	−0.184	−1.171
IBM	2.81	1.295	0.28	−1.07
TDF	3.11	1.389	−0.01	−1.329
BMT	3.54	1.343	−0.542	−0.882
CPSI	3.34	1.126	−0.47	−0.293
AT	3.09	1.529	0.0154	−1.512
CPD	3.07	1.313	0.023	−1.08
DSCM	3.09	1.551	0.033	−1.527
SSS	3.61	1.253	0.237	−1.121
ES	3.22	1.511	−0.191	−1.403
AIDSS	3.14	1.161	0.178	−0.658
SPT	2.88	1.266	0.14	−1.021

assess the performance and variability of 17 GAI functions in adaptive social manufacturing, emphasizing their roles in achieving Industry 5.0 sustainability goals. The mean values reveal the central tendency of expert evaluations, with BMT as the highest-rated function (mean = 3.54), reflecting its strong perceived importance across multiple sustainability dimensions, particularly ethical and managerial sustainability. In contrast, WETP and IBM, with lower mean scores of 3.00 and 2.81, respectively, suggest a more moderate influence in the evaluated contexts.

Variability in ratings, represented by standard deviation (SD), highlights differences in expert opinions. AT and DSCM exhibit high SD values (1.529 and 1.551, respectively), indicating diverse perspectives, possibly due to varying industry applications or levels of familiarity. Conversely, CPSI demonstrates lower variability (SD = 1.126), suggesting a stronger consensus on its impact. These variations in consensus reflect the evolving maturity and contextual relevance of different AI functions in adaptive manufacturing systems. Skewness and kurtosis further refine the analysis by indicating the shape of rating distributions. Negative skewness in functions like ROA (skewness = −0.482) suggests a concentration of higher ratings, reflecting optimism about their sustainability contributions. Conversely, positive skewness, as seen in SSS (skewness = 0.237), indicates a tendency toward lower evaluations. Most kurtosis values hover around 1, denoting moderate peakedness and balanced distributions, reinforcing the reliability of expert assessments. The analysis of individual functions reveals key trends. BMT’s high mean and moderate SD underscore its widely acknowledged role in fostering fairness, inclusivity, and ethical decision making. DSCM, with a balanced mean of 3.09 but high variability, reflects divergent perspectives, likely influenced by sector-specific challenges or varying levels of technological adoption. WETP, despite its moderate mean score, shows high variability (SD = 1.479), suggesting polarized views, potentially due to differences in organizational integration. SMD, with a mean score of 3.12 and low skewness, signifies its specialized but critical role in advancing environmental and technological sustainability through innovative materials research.

The normality tests using the Kolmogorov–Smirnov and Shapiro–Wilk methods (

Table 5. Normality tests (Kolmogorov–Smirnov and Shapiro–Wilk).

Variable	Kolmogorov–Smirnov Statistics			Shapiro–Wilk Test
	Test Statistics	df	p-Value	Test Statistics	df	p-Value
PM	3.41	1170	0	−0.058	1170	0
ROA	3.42	1170	0	−0.482	1170	0
SMD	3.12	1170	0	0.004	1170	0
WETP	3	1170	0	−0.002	1170	0
CET	3.2	1170	0	−0.124	1170	0
SCO	3.04	1170	0	−0.184	1170	0
IBM	2.81	1170	0	0.28	1170	0
TDF	3.11	1170	0	−0.01	1170	0
BMT	3.54	1170	0	−0.542	1170	0
CPSI	3.34	1170	0	−0.47	1170	0
AT	3.09	1170	0	0.0154	1170	0
CPD	3.07	1170	0	0.023	1170	0
DSCM	3.09	1170	0	0.033	1170	0
SSS	3.61	1170	0	0.237	1170	0
ES	3.22	1170	0	−0.191	1170	0
AIDSS	3.14	1170	0	0.178	1170	0
SPT	2.88	1170	0	0.14	1170	0

) revealed that the data did not strictly follow a normal distribution, as indicated by p-values consistently below 0.05 across all variables. This suggests that the assumption of perfect normality, essential for certain parametric analyses, is not fully met. However, skewness and kurtosis values, ranging between −2 and +2, indicate approximate normality, ensuring the robustness of subsequent multivariate analyses. The Kolmogorov–Smirnov test, which assesses the goodness of fit of a sample distribution against normality, consistently rejected the null hypothesis. Similarly, the Shapiro–Wilk test, particularly sensitive to departures from normality in small to medium samples, confirmed this finding. Despite these results, skewness and kurtosis analysis provided additional reassurance about the dataset’s integrity. Negative skewness in variables such as ROA and CPSI suggests a slight bias toward higher ratings, whereas positive skewness in SSS indicates a prevalence of lower scores. These trends are typical in complex, multidimensional datasets, where minor departures from symmetry are common.

The MANOVA analysis in

Table 6. MANOVA results.

Test	Measure	F	Hypothesis df	Error df	p-Value
Spillover effect	7.073	516.986	136	9216	0
Lande and Wilks	0	832.631	136	8359.751	0
Hotelling effect	106.715	897.075	136	9146	0
Largest Root Effect	28751	1948.3	17	1152	0

examines variations in the impacts of the 17 generative AI functions across nine sustainability dimensions. The results, with a Wilks’ Lambda value of 0, an F-statistic of 832.631, and a p-value below 0.05, confirm statistically significant differences among the AI functions. These findings underscore their diverse roles in achieving Industry 5.0 sustainability objectives, demonstrating tailored impacts on environmental, social, and economic goals. The low Wilks’ Lambda value highlights the strong discriminatory capability of these functions across sustainability metrics, while the F-statistic quantifies their substantial differences. The p-value further reinforces the statistical reliability of these findings, indicating that the observed variations are meaningful rather than incidental. Additional test measures, such as Hotelling’s Trace and Largest Root Effect, provide further granularity. While functions like DSCM and ES exhibit broad impacts across multiple dimensions, others like SSS and CPD show more specialized contributions. This differentiation highlights the strategic alignment of each AI function with specific sustainability goals, offering actionable insights for their prioritization and implementation. The significant MANOVA results reveal systemic interactions between AI functions and sustainability objectives. Functions such as DSCM and ROA emerge as pivotal for enhancing resource efficiency and system resilience in environmental and managerial dimensions. Similarly, BMTs significantly contributes to ethical and social sustainability by promoting fairness and inclusivity in decision making. These findings suggest that generative AI is not merely a collection of isolated technologies but an interconnected framework driving Industry 5.0’s multidimensional sustainability agenda.

Correlation analysis further supports these insights by illustrating how GAI functions contribute to different sustainability goals. Functions optimizing supply chains and resource management show strong correlations with environmental sustainability, while those enhancing workforce engagement and fairness align more with social and ethical dimensions. These results highlight the interconnected nature of GAI-driven manufacturing and emphasize the need for a holistic approach to sustainability in Industry 5.0. Moreover, the analysis captures the nuances of GAI integration into adaptive social manufacturing. While high-performing functions contribute broadly across sustainability goals, specialized functions serve niche but essential roles. For example, DSCM enhances supply chain adaptability and environmental efficiency, whereas SSSs focus on workplace safety, aligning with human sustainability. These distinctions underscore the importance of a balanced deployment strategy that integrates both versatile and specialized GAI functions for optimal collective impact.

The MANOVA findings in Table 6 provide a robust statistical foundation for understanding the differential impacts of generative GAI in social manufacturing. Correlation analysis further explores interdependencies, revealing strong positive correlations (r > 0.6) between DSCM, ROAs, and environmental/economic sustainability, while BMTs and CETs exhibit higher correlations (r > 0.5) with ethical and social sustainability. These results reinforce the evaluation findings and offer strategic guidance for aligning AI deployment with Industry 5.0 sustainability objectives. By highlighting the diverse contributions and interactions among GAI technologies, the analysis underscores the critical role of targeted AI integration in fostering resilient, inclusive, and sustainable manufacturing ecosystems.

The Friedman test results in

Table 7. The average ranking of the 17 artificial intelligence functionality variables across the 9 domains of sustainability and their prioritization.

Variable	TS		EthS		EcS		SS		EnS		CS		SCS		HS		MS
	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking	Mean	Ranking
AIDSS	13.06	4	8.41	9	12.43	4	6.03	11	7.68	11	15.02	1	8.42	11	8.79	8	2.51	14
AT	12.92	6	4.37	15	12.17	6	5.54	13	14.7	3	2.35	17	14.95	4	12.58	4	2.33	16
BMT	2.52	15	8.38	10	12.64	3	15.49	1	11.59	5	14.79	4	8.55	10	15.21	3	5.08	12
CET	2.38	17	8.51	8	7.93	10	13.03	4	7.77	10	14.9	2	5.05	14	8.13	11	14.75	3
CPD	9.09	9	1.57	17	15.17	2	9.18	10	4.3	14	14.82	3	8.56	9	4.61	14	11.6	6
CPSI	9.45	8	12.1	7	7.88	11	13	5	7.51	12	11.83	6	2.47	15	8.45	10	15.02	1
DSCM	5.33	13	15.28	3	2	16	15.4	3	2.02	16	4.88	13	12.05	5	15.32	2	8.04	10
ES	15.89	1	4.67	12	7.99	9	9.34	8	15.01	2	2.53	15	2.46	16	12.54	5	14.75	3
IBM	13.38	3	15.34	2	4.52	12	5.79	12	1.88	17	8.33	9	11.92	7	4.88	13	5.12	11
PM	15.77	2	12.26	6	8.28	8	5.41	14	11.43	7	8	11	14.98	3	5.05	12	8.29	8
ROA	13.02	5	15.38	1	4.23	14	2.52	17	7.78	9	11.28	8	11.98	6	8.46	9	14.84	2
SCO	8.87	11	4.62	13	2.11	15	9.5	7	11.37	8	11.91	5	15.07	1	2.18	16	11.77	4
SMD	9.51	7	8.36	11	15.25	1	9.2	9	4.43	13	5.23	12	2.46	16	15.58	1	11.67	5
SPT	8.88	10	4.5	14	8.44	7	2.6	16	4.27	15	11.54	7	15.03	2	4.53	15	11.58	7
SSS	2.5	16	12.53	4	4.4	13	12.79	6	11.55	6	8.27	10	5.14	13	2.12	17	5.06	13
TDF	5.35	12	12.36	5	15.25	1	2.7	15	15.02	1	4.85	14	5.23	12	12.31	6	8.16	9
WETP	5.07	14	4.35	16	12.29	5	15.47	2	14.69	4	2.45	16	8.66	8	12.26	7	2.43	15

provide a robust ranking of the 17 generative AI functions based on their median impact across Industry 5.0’s nine sustainability dimensions. As a non-parametric method, the Friedman test effectively evaluates ordinal data, identifying relative differences in AI contributions to various sustainability goals. The rankings reveal clear distinctions, with DSCM and ES consistently achieving top positions due to their significant influence, particularly in environmental and managerial sustainability. Conversely, functions such as SPT and SSS ranked lower, reflecting their more specialized roles within the broader sustainability framework. DSCM emerged as a standout performer, excelling in supply chain sustainability and environmental efficiency, where its adaptability to dynamic market conditions proved invaluable. Its top rankings across multiple dimensions underscore its systemic importance in sustainability strategies. Similarly, ES demonstrated a high impact in human and social sustainability by optimizing workplace ergonomics and fostering inclusive environments. In contrast, lower-ranked functions like SPT and SSS had narrower applications. While SPT supports long-term strategic planning and resilience building, its influence across cultural or ethical sustainability was limited. SSS, primarily focused on workplace safety, had a localized impact, contributing significantly to human sustainability but offering less value in other dimensions. These findings highlight the differentiated roles of generative AI functions, where some have broad, cross-dimensional impacts while others serve highly specialized purposes.

The Friedman test rankings also revealed key patterns among other AI functions. BMT ranked prominently in ethical and managerial sustainability, emphasizing its role in ensuring fairness and inclusivity in decision making. ROA stood out for balancing environmental and economic objectives, contributing to both efficiency and cost-effectiveness. Meanwhile, PM and CPD ranked moderately, reflecting their contributions to technological and cultural sustainability, respectively. The mean rankings and statistical measures accompanying the Friedman test results further validate these insights. For example, DSCM’s high mean ranking confirms its systemic significance, while the lower mean scores of SPT and SSS reinforce their more targeted applications. From a strategic perspective, the Friedman test results provide actionable guidance for AI deployment in adaptive social manufacturing. High-ranking functions like DSCM and ES should be prioritized for their broad-based impacts, while lower-ranked yet specialized functions such as SPT and SSS can be integrated to address specific challenges. This ensures an optimized balance between versatility and niche functionality.

Partial eta squared (η²) values were calculated to measure the effect size for each sustainability dimension. The high η² values (all above 0.7) indicate a substantial influence of sustainability dimensions on GAI effectiveness, highlighting the systemic interplay between technology and sustainability goals. The analysis identified DSCM and ES as the most impactful functions, particularly excelling in environmental and managerial sustainability. These functions significantly enhance resource efficiency, reduce waste, and improve human–machine collaboration. Moderately impactful functions, such as AT, WETP, and AIDSS, were also noted for their contributions to social and human sustainability by fostering collaboration, enhancing decision making, and supporting workforce development, critical elements of Industry 5.0. Conversely, functions such as SSS and CPD ranked lower in overall sustainability impact. While still important, they were considered less versatile compared to top-performing applications. The evaluation framework, which integrated multivariate analysis with expert assessments, effectively quantified the contributions of GAI functions to Industry 5.0’s sustainability dimensions. This comprehensive approach not only prioritized the most impactful functions but also provided insights into their interactions and roles in achieving sustainability objectives. By combining theoretical insights from content analysis and expert interviews with empirical data, the framework offers a robust, data-driven understanding of GAI’s role in sustainable manufacturing. These findings establish a foundation for the next phase of the study, which will explore the dynamic inter-relationships among AI functions using system dynamics modeling. The statistical insights gained here will inform the development of feedback loops, identify leverage points, and guide strategies for optimizing GAI implementation in adaptive social manufacturing. This ensures alignment with Industry 5.0’s overarching goals of sustainability, adaptability, and innovation.

5.4. CLD

Through three structured FGDs, 48 selected experts collaboratively mapped and validated causal relationships between the 17 GAI functions and Industry 5.0 sustainability dimensions. The first phase of the process resulted in the identification of 12 reinforcing loops (R) and 6 balancing loops (B), illustrating both positive systemic interactions and stabilizing constraints in GAI-driven sustainability applications. During the second phase, these loops were refined and validated, resulting in a final model consisting of 10 R-loops and 5 B-loops. The final phase focused on identifying high-impact leverage points, where GAI applications had the most significant systemic influence on sustainability outcomes. To assess the reliability of expert judgment, a Cohen’s Kappa inter-rater reliability test was conducted across four expert subgroups, producing values between 0.72 and 0.81, indicating substantial to near-perfect agreement. The observed agreement percentages ranged from 85% to 92%, confirming that causal relationships were consistently recognized across expert panels. The high Kappa values validate that the CLD structure accurately represents an expert consensus on the interactions between GAI functions and Industry 5.0 sustainability objectives (

Table 8. Cohen’s Kappa reliability results for CLD expert validation.

Expert Group	Participants No.	Observed Agreement (%)	Cohen’s Kappa (κ)	Standard Error (SE)	95% Confidence Interval (CI)	Interpretation
1	12	85	0.72	0.039	0.644–0.796	Substantial agreement
2	12	90	0.76	0.037	0.689–0.831	Substantial agreement
3	12	92	0.81	0.35	0.742–0.878	Almost perfect agreement
4	12	88	0.74	0.38	0.669–0.811	Substantial agreement

).

The development of the CLD, constructed using Vensim and depicted in Figure 3, provides a systemic lens through which to understand the interactions and feedback mechanisms among the 17 identified GAI functions. In contrast to the discrete statistical insights presented in Section 5.3, which quantified and ranked individual functions, this systems-based model synthesizes the complex interdependencies linking various AI applications to the multiple sustainability dimensions of Industry 5.0. By visually representing both reinforcing (R) and balancing (B) feedback loops, the CLD highlights how the integration of these functions yields emergent behaviors and dynamic trajectories that cannot be fully captured by static statistical metrics alone.

One of the key contributions of the CLD lies in its ability to reveal interdependencies and conditional relationships that were not readily apparent in numerical evaluations. For example, while DSCM and BMT emerged as high-impact functions based on expert ratings, the diagram underscores the importance of co-occurring factors. DSCM’s impact on managerial resilience, environmental efficiency, and workforce well-being becomes more pronounced when it operates in conjunction with functions that support organizational learning, employee engagement, and ethical oversight. Similarly, BMT’s role in ensuring fairness and inclusivity can trigger feedback loops that influence the pace of technological innovation and decision-making processes, fostering a balanced evolution of the system. The CLD also reveals that certain functions with moderate statistical rankings, such as IBM, serve as crucial connectors. While IBM may not have registered as a top-tier function in isolated evaluations, it emerges as pivotal within the system, facilitating the adaptation of workforce practices, strengthening market acceptance, and enabling the deployment of ROAs in ways that yield far-reaching benefits. Such relational significance is often concealed when functions are considered in isolation but becomes evident when viewed within a broader network of cause-and-effect relationships.

Additionally, the CLD provides insights into the temporal and conditional nature of sustainability improvements. Functions like SMD may not produce immediate, stand-alone enhancements; rather, their positive impact materializes over time as they interplay with other functions, such as CPD and PM, to gradually foster both environmental and cultural shifts. This temporal layering of effects, unavailable through one-off statistical snapshots, illustrates how long-term sustainability is an emergent property of an evolving, interconnected system rather than the sum of discrete interventions. Moreover, the diagram places the observed variability and dispersion in expert opinions (identified in Section 5.3) into a more meaningful context. High variability in ratings for AT, for instance, can be reframed as stemming not solely from differences in familiarity or sectorial applications but from the complexity of AT’s position within a dense web of interdependent loops. Different experts may have implicitly accounted for a variety of indirect effects, feedback mechanisms, and interaction sequences, reflecting the underlying systemic complexity that the CLD makes explicit.

The CLD’s systems-based perspective augments and complements the statistical results. It illuminates hidden relationships, identifies leverage points, and clarifies how certain functions gain or lose influence within a networked ecosystem. By focusing on dynamic feedback rather than static comparisons, the CLD offers guidance for targeted interventions, policy formulations, and strategic planning that align more closely with the holistic vision of Industry 5.0 sustainability. The resulting insights underscore the necessity of viewing GAI functions as co-evolving components of a complex system, rather than as isolated tools, and lay the groundwork for more nuanced, adaptive, and integrated approaches to achieving sustainability goals in the manufacturing sector.

6. Discussion

The findings of this study underscore the transformative potential of GAI in adaptive social manufacturing, aligning closely with Industry 5.0’s vision for a human-centric, sustainable, and resilient industrial ecosystem. Through the integration of empirical analysis, expert validation, and system dynamics modeling, this study presents a robust and multidimensional framework that extends the current understanding of AI-driven transformations in manufacturing across environmental, economic, social, and ethical dimensions. At the core of this study is the recognition that GAI, with its generative capabilities to process, simulate, and create, acts not only as a technological enabler but also as a catalyst for systemic change. Technologies such as DSCM and ROAs have demonstrated significant capacity to simulate complex scenarios and provide real-time decision support. These functions directly contribute to enhancing operational agility, mitigating disruptions, and improving system resilience, capabilities that are particularly critical in the face of fluctuating market demands and global supply chain uncertainties [53].

From an environmental perspective, tools like PM and SMD play pivotal roles in minimizing material waste, reducing energy consumption, and improving resource efficiency. These functions directly support circular economy practices by enabling resource reuse and material recycling [78]. The integration of GAI tools enables manufacturers to transition from linear production models to more sustainable, regenerative systems aligned with the goals of Industry 5.0 and the broader SDGs. Economically, GAI empowers the creation of flexible, innovation-driven business models that center on mass customization, cost optimization, and consumer responsiveness. Technologies such as CPDs and AIDSSs facilitate rapid adaptation to changing consumer needs while optimizing operational costs [36]. Moreover, platforms such as WETPs enhance workforce adaptability and learning, preparing human capital for the demands of AI-integrated industrial environments [33]. This combination of efficiency and empowerment is essential for achieving long-term economic sustainability and competitiveness.

On the social dimension, GAI’s potential to enable inclusive and participatory manufacturing environments is highlighted through tools such as CETs. These platforms support co-design and collaborative production practices that integrate local knowledge and cultural values, thereby reinforcing cultural sustainability and social legitimacy [60,61]. As the industry moves toward more decentralized, networked models of production, these social capabilities become increasingly central to legitimacy and long-term societal value creation. Ethical governance remains a central concern in the deployment of GAI systems. Tools such as BMTs and TDFs address the risks of algorithmic bias, fairness, and accountability. Their inclusion not only promotes equitable AI practices but also strengthens stakeholder trust, a precondition for societal acceptance and sustainable AI integration [37,39]. Furthermore, ensuring data privacy, transparency, and explainability in AI systems reinforces the ethical underpinnings of Industry 5.0.

The system dynamics modeling employed in this study, especially the development of the CLD, provided crucial insights into the reinforcing and balancing feedback loops among GAI functions and sustainability dimensions. These models illustrate how certain AI functions, such as DSCM and BMT, serve as key leverage points with cross-dimensional impacts. For example, while optimization tools improve efficiency, their over-dependence may introduce vulnerabilities and bottlenecks, underscoring the need for balanced and context-aware GAI strategies. Such insights mirror findings from other research that highlight the complexity of managing technological interdependencies within socio-technical systems [53]. Despite the demonstrated benefits of GAI, several barriers impede its widespread implementation. These include limited awareness, particularly among SMEs, high capital requirements, lack of skilled personnel, and concerns over data integration and infrastructure readiness. These challenges align with prior findings in the literature [55,80] and call for structured, phased adoption strategies. Policy interventions, such as subsidies for technology adoption, public–private knowledge transfer platforms, and skill development initiatives, can significantly ease the transition toward AI-enabled sustainable manufacturing.

Another significant challenge lies in addressing the ethical and regulatory implications of GAI. The governance of GAI-generated content, data usage rights, and accountability mechanisms remains underdeveloped. Effective implementation requires multi-stakeholder collaboration across industry, academia, and government to develop robust frameworks that guide ethical GAI deployment while ensuring that innovation is not stifled [40]. The broader implications of this research highlight GAI’s potential to harmonize technological advancement with sustainable development imperatives. By embedding GAI within social manufacturing systems, industries can drive transformations that align environmental conservation, economic growth, and social equity. The systemic nature of this alignment suggests that GAI should be conceptualized not as a collection of discrete tools but as a cohesive framework that links strategy, operations, and ethics. This holistic understanding distinguishes this study from prior works, which often focus solely on the technical or efficiency-driven aspects of GAI. While this research offers substantial insights, it is not without limitations. The empirical data are drawn from expert panels based in Australia, which, while globally informed, may not fully capture geographic, cultural, and regulatory diversity. Future research should replicate and extend this study across different industrial sectors and regions to validate and enrich the model. Longitudinal studies are also necessary to examine the evolving impact of GAI tools over time, especially in the light of emerging regulatory landscapes and shifting societal expectations.

In conclusion, this study demonstrates that, when thoughtfully and ethically integrated, GAI can revolutionize adaptive social manufacturing by fostering sustainability, inclusivity, and resilience. By bridging the gap between technological innovation and human-centric values, GAI offers a roadmap for a new industrial paradigm, one in which equity, transparency, and sustainability are not peripheral considerations but central tenets. This shift represents not only a technical evolution but a reimagining of the role of industry in shaping a just and sustainable future.

7. Conclusions

This study was driven by the urgent need to address challenges in modern manufacturing as it transitions to Industry 5.0. While Industry 4.0 advancements laid the groundwork for automation and connectivity, they often neglected human-centric and sustainability principles. Existing research highlights significant gaps in integrating ethical, social, and environmental considerations into industrial ecosystems. Moreover, despite its transformative potential, the application of GAI in manufacturing remains limited due to technical complexity, ethical dilemmas, and the absence of robust assessment frameworks. In response, this study explored the intersection of GAI and adaptive social manufacturing to bridge these gaps. By aligning advanced AI functionalities with Industry 5.0’s sustainability dimensions, the research aimed to provide actionable insights into how manufacturing systems can meet ethical, environmental, and social objectives. The novelty of this research lies in its integrated, systemic approach. Unlike prior studies that focused on either GAI or sustainability in isolation, this study uniquely combines GAI technologies with system dynamics modeling to develop a comprehensive framework for adaptive social manufacturing. It identifies key GAI applications, such as DSCM, ROAs, and BMTs, and examines their inter-relationships and systemic impacts through CLDs. This innovative approach deepens the understanding of how these technologies interact to drive sustainability outcomes across environmental, economic, and social dimensions.

Methodologically, the study followed a multi-stage research design combining content analysis, expert engagement, statistical evaluation, and system dynamics modeling. The content analysis identified nine critical sustainability dimensions for Industry 5.0, including environmental, social, and cultural sustainability. Expert interviews provided domain-specific insights, prioritizing 17 GAI functions relevant to adaptive social manufacturing. The statistical evaluation, incorporating MANOVA and Friedman tests, quantified the impact of these AI functions, highlighting their contributions to sustainability goals. Finally, system dynamics modeling visualized the complex interactions among AI functions, offering actionable insights into their systemic implications and optimization potential. The findings confirm that GAI plays a transformative role in advancing Industry 5.0’s sustainability objectives. Technologies like DSCM and ROAs enhance adaptability and efficiency, while tools such as BMTs and CETs address ethical and social challenges. Integrating these functions within adaptive social manufacturing enables industries to shift from traditional efficiency-driven models to inclusive, human-centric, and sustainable ecosystems. Furthermore, CLDs highlight critical feedback mechanisms, reinforcing the importance of a systemic approach to GAI integration. This study underscores GAI’s potential as a cornerstone of Industry 5.0. By bridging technological innovation with ethical and human-centric principles, it offers a pathway for industries to achieve resilience, adaptability, and sustainability. However, realizing this vision requires addressing key barriers, including ethical governance, workforce training, and resource allocation. Future research should explore the application of this framework across diverse industrial contexts to assess long-term impacts and scalability. Ultimately, the insights from this study pave the way for a more inclusive, sustainable, and innovative future in global manufacturing. To distill the primary contributions and implications of this study, the following key highlights encapsulate the core findings, methodological innovations, and strategic insights for advancing Industry 5.0 through responsible GAI integration:

• This research proposes a novel framework integrating GAI into adaptive social manufacturing systems, addressing current theoretical and practical gaps in Industry 5.0’s sustainability agenda.
• A rigorous, multi-stage methodology was implemented, comprising content analysis, multi-round expert validation (n = 130), MANOVA and Friedman tests, and causal system dynamics modeling to ensure methodological robustness and empirical depth.
• The study identifies and prioritizes 17 high-impact GAI functions based on their contributions to nine validated sustainability dimensions, including environmental, economic, social, human, cultural, ethical, and managerial domains.
• Quantitative analysis reveals that functions such as DSCM, BMT, and ES consistently scored highest in terms of sustainability impact and strategic relevance.
• Through expert-driven construction of a CLD, the research uncovers reinforcing and balancing feedback loops among GAI functions and sustainability outcomes, illustrating the systemic nature of GAI-enabled sustainability.
• The findings demonstrate that leveraging GAI within adaptive manufacturing ecosystems requires a holistic approach, balancing technological advancement with ethical governance, human-centric design, and organizational resilience.
• By bridging theoretical constructs and real-world application, this study delivers a validated, scalable, and transferable roadmap for industries seeking to operationalize Industry 5.0 through responsible and sustainable GAI deployment.