When “Human-Centered” Requires Knowing What Human Actually Means

Every conference this year has a session on AI in manufacturing. Few ask a more basic question first: if a system is going to be “human-centered,” do we actually understand what a human is?

That question matters more than it sounds. Manufacturing has spent the last decade building smarter machines, tighter data loops and more predictive systems — Industry 4.0’s promise of connected, optimized operations. But somewhere in that optimization, a critique started surfacing across the industry: technology was being prioritized over people, profitability over well-being, efficiency over the humanistic values that actually make organizations resilient. Industry 5.0 emerged in direct response — not to reject the technology but to insist that people belong back at the center of it.

That distinction is worth sitting with because it’s easy to conflate “smart” with “human.” They aren’t the same thing. An AI system can process, predict and optimize — but it doesn’t have embodied experience, moral judgment or lived history. It doesn’t carry the weight of a decision the way a person does. Philosophers have spent decades debating consciousness, and one useful frame that keeps surfacing is the idea that human beings run on two distinct modes of thought: a rule-based, process-oriented mode that craves certainty and control and an intuitive, experience-based mode that tolerates ambiguity and sees the whole picture. Skill acquisition research shows the same pattern — novices lean entirely on rules, and only through repeated mistakes and agency do they develop the intuition that marks true expertise. That leap from rule-following to intuitive judgment isn’t a data problem. It’s a human one.

This has direct implications for how leaders roll out any new system, technology or process. Three things tend to determine whether people actually adopt something new: whether they trust it, whether they retain some control over the outcome and whether it reduces or adds to their cognitive load. Get those wrong, and you get what researchers call algorithm aversion — otherwise capable people quietly refusing to use a system, not because it’s technically flawed but because it violated one of those three conditions. That aversion isn’t stubbornness. It’s a rational response to feeling replaced rather than supported.

The leadership implication is straightforward, even if it’s not easy: decision-making authority should be pushed as close to the people doing the work as possible. Systems should be built to augment judgment, not override it. And rollout timelines should be built around how people actually absorb change, not how fast a system can technically deploy.

None of this is an argument against new tools or new technology. It’s an argument for sequencing. Before asking “what can this system do,” a leader is better served asking “what does the person using it need in order to trust it, stay in control of it, and not be overwhelmed by it.” That question doesn’t expire. It will be just as relevant the next time a new platform, a new machine, or a new AI tool shows up on the shop floor — because the tools will keep changing, but what makes people willing to work alongside them won’t.

About the Author

Ryan Pohl is the founder of Praeco Skills LLC. He may be reached or ryan@praecoskills.com.

Praeco Skills LLC is an MMA Premium Associate Member and has been an MMA member company since May 2023. Visit online: praecoskills.com.

Ryan Pohl was a speaker at the 2026 MMA Workforce Solutions Summit. Learn more about MMA events.