Smarter Factories, Bigger Risks: Why AI Governance Is Becoming Industrial Policy

Artificial intelligence is rapidly moving from experimental technology to the operating core of factories, logistics networks, energy systems and industrial decision-making. However, the next phase of industrial transformation will not be determined by automation alone.

The review "Artificial Intelligence Governance and Sustainable Industrial Transformation in the Era of Industry 5.0," published in Sustainability by David Mhlanga of the University of South Africa, argues that the decisive challenge is building governance systems capable of making AI productive, sustainable, resilient and genuinely human-centred.

The study reframes the transition from Industry 4.0 to Industry 5.0 as more than a technological upgrade. Industry 4.0 largely prioritised automation, efficiency and digitally connected production, while Industry 5.0 places greater emphasis on human well-being, environmental responsibility and resilience. The shift means governments and businesses must judge AI not simply by how much productivity it generates, but by how it affects workers, resource use, industrial security, inequality and long-term sustainability.

Based on an integrative review of 63 academic and policy sources, Mhlanga proposes a Human-Centric AI Governance Framework that connects intelligent technologies with workforce empowerment, ethical oversight, cybersecurity, sustainability and the Sustainable Development Goals. The framework is conceptual rather than empirically tested, but its central message is difficult to ignore: without governance, the same AI systems that strengthen efficiency and resilience could deepen surveillance, labour disruption, environmental pressure and technological inequality.

The Industrial AI Race Is Moving Beyond Efficiency

The technological case for industrial AI is already strong. Machine learning can predict equipment failures before breakdowns occur, digital twins can simulate production systems, intelligent logistics can improve supply-chain planning, and AI-assisted energy management can reduce waste and optimize industrial consumption. These capabilities make AI central to the emerging model of adaptive, data-driven and increasingly autonomous industrial production.

However, Industry 5.0 introduces a different test of progress. An industrial system cannot be considered successful simply because it produces more at lower cost if it simultaneously erodes worker autonomy, increases exclusion or consumes unsustainable amounts of energy. Human-centricity, sustainability and resilience therefore become complementary measures of industrial performance rather than social concerns added after technological deployment.

This changes the logic of investment. Businesses evaluating AI systems increasingly need to consider not only productivity gains, but cybersecurity exposure, explainability, workforce implications and environmental costs over the technology lifecycle. Governments face a similar shift: industrial policy and AI policy can no longer remain separate domains when algorithmic systems directly influence production, employment, infrastructure and competitiveness.

The review identifies six recurring priorities across the literature: human-centred industrial transformation, environmental responsibility, industrial resilience, ethical AI governance, cybersecurity and digital trust, and workforce skills. Their importance lies in their interdependence. A highly efficient AI-enabled factory that is vulnerable to cyberattack, dependent on opaque algorithms or unable to retain skilled workers may be technologically advanced but institutionally fragile.

AI's Sustainability Promise Comes With Its Own Environmental and Social Costs

According to the study, AI can simultaneously support and undermine sustainability. Intelligent systems can improve energy efficiency, monitor emissions, reduce material waste and enable circular-economy practices. Predictive maintenance can extend machinery life, while smarter logistics and production planning can reduce unnecessary resource consumption.

AI depends on computational infrastructure with substantial energy demands. Data centres, high-performance computing and continuous data processing consume electricity and can increase carbon emissions. The review therefore challenges the assumption that digitalisation is inherently green: the environmental gains created by AI applications must be weighed against the footprint of the infrastructure required to operate them.

The social trade-offs are equally significant. AI-assisted systems can improve occupational safety, remove workers from dangerous tasks and strengthen human-machine collaboration. Yet poorly governed automation can also displace jobs, intensify workplace surveillance and widen inequality between workers who possess advanced digital skills and those whose roles are most vulnerable to technological substitution.

For developing economies, this tension is particularly consequential. AI-enabled industrialisation could help countries leapfrog older production models, improve resource efficiency and participate in higher-value global supply chains. But weak digital infrastructure, limited regulatory capacity and skills shortages could also concentrate the benefits among already advantaged firms and regions, making technological adoption another source of inequality rather than a broad-based development strategy.

The Real Bottleneck May Be Governance, Not Technology

The review argues that existing AI governance approaches remain fragmented. International frameworks have already established principles around transparency, accountability, fairness, safety and human-centred values, while Industry 5.0 literature emphasizes sustainability and resilience. What remains underdeveloped is an integrated model connecting those principles directly to industrial operations, worker participation, environmental performance and technological resilience.

The proposed framework attempts to close that gap by linking four broad areas: intelligent technologies, human-centred industrial processes, governance mechanisms and sustainability outcomes. Technologies such as AI, robotics, IoT and digital twins are therefore treated as only one layer of transformation. Workforce empowerment, digital literacy, transparency, accountability, cybersecurity, responsible innovation and environmental performance are positioned as equally necessary components.

This integrated approach has significant policy implications. Governments need regulatory capacity capable of examining algorithmic decisions, protecting industrial data, setting cybersecurity requirements and assessing environmental impacts. Companies, meanwhile, need governance structures that establish who remains accountable when AI systems influence hiring, production, worker monitoring, safety or allocation of industrial resources.

Regulatory fragmentation presents an additional risk because industrial AI operates across borders, supply chains and digital infrastructures. Divergent rules on data, cybersecurity and algorithmic accountability could increase compliance costs and create gaps that companies or malicious actors exploit. Greater international coordination is therefore likely to become part of industrial competitiveness itself, not merely a matter of technology diplomacy.

Industry 5.0 Will Be Won Through Skills, Trust and Institutional Capacity

The future of advanced manufacturing will depend as much on institutions and people as on algorithms. Industry 5.0 assumes that AI should augment human capabilities rather than simply remove labour from production. That requires sustained investment in reskilling, digital literacy, lifelong learning and organisational adaptation.

Educational systems thus become a vital component of industrial strategy. Universities and training institutions will need to combine technical skills with knowledge of ethics, sustainability, cybersecurity, law and governance. The workforce required for Industry 5.0 will not consist only of programmers and engineers, but also professionals capable of understanding how intelligent technologies affect institutions, workplaces and societies.

The framework also connects responsible industrial AI to SDG 8 on decent work, SDG 9 on industry and infrastructure, SDG 11 on sustainable communities, SDG 12 on responsible production and SDG 13 on climate action. This connection is strategically important because it places AI governance within development policy rather than treating it as a narrow technology-regulation issue.

The study should nevertheless be interpreted carefully. It is an integrative conceptual review rather than an empirical evaluation of real-world industrial governance. The 63-source evidence base was not formally scored using a standardized quality-assessment tool, only English-language sources were included, and a single author conducted study selection, extraction and coding, creating potential selection and interpretive bias.

Most importantly, the proposed governance framework has not yet been validated in factories, supply chains or regulatory systems. Future research needs expert panels, case studies, surveys and sector-specific testing across manufacturing, energy, logistics and healthcare, particularly in developing economies where institutional and infrastructure conditions may differ substantially from those of advanced industrial states.