By Shin Nakamura, President of one to ONE Holdings
Manufacturing has been one of the most willing industries to embrace AI, and the testament to that spans decades. What’s changed is the pace of adoption: Industry 4.0 is well underway, and demands on manufacturers have reached unprecedented heights. That backdrop explains the accelerated enthusiasm for more sophisticated, increasingly autonomous AI tools and systems, from China and Japan’s early moves into robotics and shop-floor automation to India, where 97% of manufacturers now call digital transformation “essential” to the industry’s future.
Yet there’s a massive gap between promising to adopt AI and actually scaling it on the factory line, and that chasm is growing. The usual explanation is a lack of workforce readiness or data gaps that keep AI from being safely integrated at scale. That diagnosis misses the real story. The gap isn’t because workers resist AI, but because leadership skips important groundwork needed to make AI viable on the factory floor in the first place.
Worker Resistance is a Misdiagnosis of a Leadership Problem
New technology almost always meets resistance, and AI is no exception. Mobile devices, proprietary software, and computers were all once treated as unwelcome incumbents until workers understood why they’d been brought into the workflow. Building that understanding starts at the top, and it’s too often the task that gets skipped. People resist what they don’t understand, so leaders need to show, not just state, the purpose behind an AI tool.
The deeper problem is that this understanding gap runs both ways. Leadership and administration are often removed from the operational day-to-day, and workers know it: 77% of frontline workers say leadership doesn’t understand daily operations. That distance has a direct cost. Only 3% of AI frontline recommendations get implemented, and leaders lose visibility into who is actually working the line and how technologically fluent they are, so training never lands on target.
That disconnect also obscures real, and different, needs across the workforce. Generational attitudes toward technology vary widely: 83% of Gen Z and 73% of Millennials already use AI at work, versus 60% of Gen X and 52% of Baby Boomers. That’s not a resistance problem so much as a design problem. In my own experience, meeting these differences means applying different interfaces accordingly, such as voice-input tools paired with hands-on training that makes the “why” of each tool explicit. Build the interface around the user, not just the use case.
Trust is the crucial foundation to all of this. Leaders are responsible for making both the rationale for a tool and the training to use it transparent and accessible. Skip that, and adoption stalls before it starts. Without that trust, only 35% of individual contributors will embrace AI at work.
Where AI Initiatives Actually Fail
AI development is moving extremely fast, and the instinct to rush to deployment, especially when competitors are seemingly racing ahead, is understandable. But it’s also why leaders often make the call, hand it down, and skip the groundwork.
Before deploying any AI system or tool, leadership must set the right priorities—in the right order. Every AI rollout in manufacturing should be anchored to a clear order of priority: safety first, quality of production next, productivity gains after that. This sets a clear directive that enshrines workers’ wellbeing at the core—an important pillar for building trust and remaining clear on operational and business objectives.
Another leadership mistake is not asking the floor before redesigning workflows with AI in the mix. A recent Gartner survey of CHROs found that 78% agree workflows need to change to get value from AI. But redesigning a workflow requires first confirming, in detail and from the shop-floor perspective, how it actually works today.
Workflow design in a manufacturing setting can’t be rooted in theory but in what happens on the line. That means leaders need to ask frontline workers directly: what are the existing pain points? What’s causing operational friction? Only once those answers are in hand should a new workflow be proposed, explicitly tied to the problems it solves. That added clarity also builds trust and willingness to engage.
The Organisational Changes That Determine Success
Getting this right requires the right people in the room before AI reaches the factory floor and production line. There are three perspectives that are absolutely necessary for a structural safeguard for AI in manufacturing, on any proposed change:
Operators, who catch practical, ground-level issues to secure safety, fulfill quality, and improve productivity in their operations that are often overlooked in boardroom discussions.
Engineers, who bring the technical view to secure safety, fulfill quality, and improve productivity by using machines and equipment, including how the AI tool integrates with existing machinery, data flows, and process logic.
Quality control personnel who anchor the change to secure safety, fulfill quality, and improve productivity for processing materials and finishing products by making sure speed gains do not come at the cost of defects or risk.
Supporting teams should also do the preparatory work before asking floor workers for input. Floor staff already carry a full daily workload, so engineering- and QC-led teams should map out the proposed workflow in detail first: what’s changing, why, and what the expected benefit is. Operators can then react to a concrete proposal instead of starting from a blank page, which reduces the stress and confusion common in early-stage rollouts, when the system itself is often still unreliable for lack of data or experience.
The goal is to put the existing workflow and the proposed workflow side by side and have each function weigh in on both. Friction points get flagged and addressed before rollout, whether that’s a step that works for engineering but creates an operational bottleneck, or a change that speeds things up but weakens a quality checkpoint or puts a worker’s safety on the line. That structure is what lets leadership evaluate real trade-offs and further justify the “why” behind its decisions.
Building Ownership, Not Just Compliance
This is also builds ownership. The floor doesn’t need to design the new system, but its feedback should be reflected in the final version. When workers can see that their input changed something, they feel ownership over the process rather than having a decision imposed on them.
It’s worth noting that traditional software rollouts arguably did not require this rigor to the same degree because those systems are more fixed in nature. AI systems, particularly more autonomous ones, are more fluid, more detailed, and more complicated. They adapt based on inputs, data quality, and usage patterns, meaning that cross-functional review is absolutely vital for AI rollouts.
Building Clear Pathways for Shared Knowledge and Collaboration
Many manufacturers do not have a standardized glossary or knowledge bank that is accessible across the factory floor and administrative departments, including higher-up leadership. Those barriers to sharing knowledge and clear communication not only distance leadership from the shop floor, but also evolve into safety threats, limited training, and technical breakdowns that make workers more wary of new technologies.
Moreover, inconsistent inputs are a key cause of errors and hallucinations in AI systems, and workers often end up blamed for the faulty outputs that result. Alongside cross-functional review, manufacturers need clear pathways for communication and knowledge sharing that keep the “why” behind AI embedded in everyday practice, starting with a shared glossary and clear input procedures built into the rollout itself.
Leaders should treat this the way they’d treat onboarding a new employee: it’s not reasonable to expect anyone, human or algorithm, to distinguish right from wrong without clear standards. Ensuring that there is a constantly accessible—and updated—knowledge bank and glossary not only eases training for new hires. It also supports upskilling and familiarity among existing staff, providing a reliable point of reference at all times, preventing hallucinations and confusion, and accelerating stable and effective learning of AI. This can also be used as an indicator for redesigning workflows—what current protocols or procedures are, what might be missing, what could be causing problems, and more.
The gap between the boardroom and the shop floor is arguably most felt with AI deployment. The answer is not simply a training fix or a mindset change, but also a carefully planned, orchestrated, and collaborative procedural improvement championed by leadership. Ultimately, leadership must serve the workforce by helping them integrate tools to improve their roles, rather than using an isolated top-down command structure that is removed from operational and production realities.