The Evolution of AI Workforce in Manufacturing

Discussions around generative AI have largely focused on how intelligent these systems have become. However, for manufacturing, the critical question is not how sophisticated AI outputs are, but how reliably AI can perform work on an ongoing basis. What is changing is not simply the capability of AI, but its role—AI is beginning to move from a tool that supports work to an entity that actively performs it within the enterprise.

Historically, AI in manufacturing has been applied to optimize specific functions such as demand forecasting and quality control. While generative AI has significantly enhanced knowledge work, most use cases still remain limited to content generation and conversational support. More recently, however, a new class of AI has emerged—systems capable of understanding goals, planning tasks, and executing workflows autonomously through interaction with external tools. This represents not just an improvement in performance, but a fundamental shift in how intelligence is applied in business operations.

Manufacturing is uniquely positioned in this transition, as it spans both digital and physical domains. As a result, this evolution is not confined to white-collar work but is increasingly extending to shop-floor operations. AI is no longer just a support tool; it is evolving into an executional entity with defined roles, capable of continuously performing tasks.

In this context, the key question is no longer where to apply AI, but how to redesign the workforce itself—on the assumption that humans and AI will work side by side. Manufacturing is now moving from isolated automation toward a holistic redesign of how work gets done.

With the advancement of generative AI, the role of AI shifts from a tool that supports work to an entity that actively performs it. This shift is particularly relevant for manufacturing, where complex value chains—spanning design, procurement, production, maintenance, quality, and service—create a natural fit for AI systems capable of operating autonomously across multiple functions.

To understand this transition, the concept of the “AI workforce” becomes critical. Rather than viewing AI as isolated automation tools, it is more useful to define AI as a form of labor embedded within the organization, performing multiple tasks under defined roles and collaborating with humans on an ongoing basis. In this context, AI workers can be broadly categorized into “digital workers,” which handle white-collar tasks, and “physical AI workers,” which operate in real-world environments (see Figure).

While digital workers primarily operate in virtual environments, manufacturing is now entering a phase where AI begins to act directly on the physical world through robots and industrial systems. By sensing conditions, making decisions, and executing actions via actuators, AI is increasingly able to perform real-world tasks autonomously. This convergence of cyber and physical systems is bringing “physical AI” from concept to reality.

This shift is driven by the convergence of advances in robotics and AI. Traditional industrial robots were designed for repetitive, pre-defined tasks, but next-generation systems are evolving toward adaptive operations capable of perceiving environments, making decisions, planning actions, and executing them in dynamic conditions. In particular, embodied AI, including humanoid robots, is attracting growing attention as a potential platform for more generalized physical work, with global competition accelerating across regions and industries.

As discussed in the previous chapters, AI is expanding its reach from digital tasks into physical operations, fundamentally reshaping the workforce in manufacturing. The key shift is that AI adoption is no longer about deploying isolated tools, but about redesigning the workforce itself on the assumption that humans and AI will work side by side. The question is no longer whether to use AI, but where to introduce it, how to define roles, and how deeply it should be embedded into operations.

First, organizations must move beyond “adding AI” to existing processes and instead rethink the workflow itself. A phased approach—starting with digital workers and gradually extending toward integrated models that combine digital and physical AI—will be critical to building end-to-end execution capabilities.

Second, it is essential to distinguish and integrate two dimensions of deployment: digital as organizational integration and physical as operational integration. Success in the former depends on role definition, KPIs, governance, and management processes, while success in the latter hinges on safety, quality, reliability, and alignment with existing factory systems.

Third, the expansion of AI does not diminish the role of humans; it elevates it. Organizations will need frontline capabilities to effectively collaborate with AI, managerial capabilities to supervise and intervene when necessary, and leadership capabilities to design and optimize the overall human–AI system. At the same time, ultimate responsibility for safety, quality, and accountability remains with humans.

Looking ahead, the ability to leverage AI workers while balancing performance with safety and trust will become a core capability in manufacturing. This is not simply a technological transition—it is a transformation of how work itself is designed and executed.