Across industries, 63% of employers say the skills gap is the main barrier to transforming their business, and 77% plan to train their people to work alongside AI, according to the World Economic Forum’s “Future of Jobs Report 2025.”
In regulated manufacturing, AI tools are arriving faster than people are developing the skills and habits needed to use them safely. The risks, ways of working and skills gaps also change depending on where in the business an AI tool is deployed. At LaSalle Group, we see another risk beyond implementation and adoption: People may not only refuse to use AI, but some may also trust it too much. Overreliance erodes the critical thinking skills and the technical and regulatory knowledge teams build over time, and it creates risks that human-in-the-loop controls may not catch.
This article explains what AI workforce development is, where the real gap lies on the plant floor, what must remain a human responsibility in a GxP environment and how LaSalle prepares teams to work with AI.
What is AI workforce development?
AI workforce development is the structured process of preparing employees to work effectively and responsibly with artificial intelligence. It includes:
- Assessing current skills
- Redesigning roles and workflows
- Delivering hands-on training
- Managing change to build the right level of trust in AI tools
“When talking about AI workforce development, just like in any transformation, you need to understand what the baseline status for your organization is,” said Liz Meléndez, head of LaSalle’s Talent Development & Innovation Center (TDIC). “With AI, that baseline is going to have variability. You’ll have early adopters and, in regulated manufacturing, a big group of laggard adopters.”
Where AI workforce development differs from traditional workforce development is in risk mitigation and how processes and roles are redesigned. “AI opens a world of new opportunities. In digital transformation, we optimized processes, standardized and digitalized,” said David Pagan, CEO of Murmuratto and chief innovation officer of LaSalle Group. “With AI, we can rethink that process entirely, and with the right guardrails and observability, the bulk of the work can be done and delivered to a human for approval.”
Pagan frames every use case with two questions: Is it cheaper to execute the work with a person or with an AI agent? And which is more reliable? He suggested that in the context of regulated manufacturing, there is a third and equally important question: What is the risk involved?
If AI does not deliver a tangible return on the work, we need to ask ourselves: Am I using the wrong tool? Does the original process or way of working still make sense? Should I even apply AI to this function? “Quite honestly, there are some things, like writing a short email, that take less time for me to write and send than to figure out a prompt or revise what the agent developed for me,” Pagan said. “For example, let’s push this meeting back five minutes.”
He also points to a practical problem: AI technology moves so fast that building a curriculum around new tools is a risky bet. By the time a course is finished, it may already be outdated. That is why AI workforce training should focus on skills that outlast any single tool.
AI workforce development should not be a catalog of generic AI courses or a tool rollout. It should be role-specific and built around the work an employee is expected to execute. A workforce development initiative also cannot stand alone.
It should not be an afterthought. It should also not become a priority only when strategy fails or when a new AI tool deployment fails, stalls, puts operations at risk or, worse, generates risk within regulatory compliance.
Getting value from AI takes a coordinated strategy that spans the organization’s growth and innovation roadmap, governance, deployment and pilots, capability building, regulatory compliance, and ongoing monitoring and risk mitigation, supported by sound change management. AI workforce development is the foundation of that strategy.
The AI gap in regulated manufacturing
Human capital has the lowest maturity of all smart manufacturing areas, and only 29% of manufacturers use AI or machine learning at the facility or network level, according to Deloitte’s “2025 Smart Manufacturing Survey.” The data points to people as the weakest link, and LaSalle’s experts see the same pattern on the plant floor.
Two groups on the plant floor
- Enthusiastic adopters see AI as a tool for agility and process efficiency.
- Skeptics fear job displacement or errors in AI-generated outputs.
The trust gap cuts both ways
The core problem is trust calibration: knowing how much to rely on what an AI tool produces. People either do not know how much to believe AI outputs, or they trust them blindly without checking for accuracy. However capable the models have become, there is not always a verifiable source behind what a large language model produces.
Pagan describes the overreliance curve this way: The first time you use AI for a task, you review the output carefully. The second time, you review it again. The third time, you feel confident. The fourth time, you approve it without reading it.
What must remain a human responsibility
In a GxP environment, AI can change how work is done, but not who is accountable for it. Not every individual AI output needs a person to review it. A validated, static and deterministic model working within defined limits, such as in-line quality inspection, can run without a review of each result. Generative AI is different: Its outputs must be reviewed and cleared by an authorized person before they are used. In every case, the quality unit remains accountable for each batch.
Where regulatory matters are involved, the use of AI is restricted and limited. That makes the controls around how AI is used, and compliance with the requirements that govern it, as important as the tools themselves.
That is why the overreliance curve matters. A review step only protects the process if people actually review. Human-in-the-loop controls work when the human in the loop is trained to question the output, not just approve it.
The foundational skills for AI success
Because tools change faster than any curriculum, the better investment is in skills that stay useful as models evolve and new ones go live. In regulated manufacturing, that starts with decision-making, risk management and contextual judgment.
There is also a challenge for people early in their careers. AI is here to help people work with more agility, but it should not replace critical thinking and self-reflection. Early-career employees in particular may be tempted to take the path of least resistance and skip the reasoning process. That reasoning is how they build knowledge over time. LaSalle’s answer is a simple rule: Think first, then use AI. Frame the problem and reason through it before turning to a tool.
Gartner identifies six capabilities in two core areas: work-orchestration skills and oversight and governance skills. Focusing on these foundational skills builds a workforce that can evolve with AI.
Work-orchestration skills
The first set of skills is grouped under the umbrella of work-orchestration skills.
Problem formulation
An AI tool’s output is only as good as the input it receives. Problem formulation is the ability to scope a problem, set objectives and define boundaries before asking AI for help. In a GMP plant, that means framing a deviation, a trend or a batch record question precisely before any tool gets involved.
Systems thinking
As AI reengineers workflows, people need to understand how individual components connect within the larger whole. A July 2025 Gartner survey of 3,029 employees found that systems thinking improves employee performance by 1.6 times in AI-enabled environments. In regulated manufacturing, a change at one step can ripple into validation, documentation and downstream quality.
Strategic foresight
As AI matures, it will enable entirely new ways of working. Strategic foresight is the ability to anticipate those shifts and prepare roles and processes for them, instead of reacting after the fact.
Oversight and governance skills
The second set of skills is grouped under the umbrella of oversight and governance skills and includes:
- AI judgment and risk mitigation
- Human-AI mediation
- Ethical leadership
How to prepare your team: The LaSalle approach
If we were on site with your leadership team today, these would be the key points of our first conversation:
- Understand the company’s strategy and key performance indicators.
- Understand the company’s AI and digital transformation roadmap.
- Assess the baseline.
- Prioritize focus. Identify where AI can bring value to your team and your customers.
- Prioritize roles where AI will change the work most.
- Train on the work your team is expected to deliver with the principles above, developing use cases and fine-tuning models and applications together with the team.
- Measure adoption and the quality of decisions, not just course completion. AI workforce readiness shows up in how people work, not in a training record.
What we learned integrating AI into our courses
LaSalle has integrated AI into its ready-to-deploy courses, which include Lean Six Sigma, Project Management, Problem Solving and Process Knowledge Robustness. In fact, AI is now part of every LaSalle course.
The baseline survey showed most of our team open to AI as a productivity tool, and the courses apply the “think first, then use AI” rule described above. An early, practical sign: Participants have been submitting coursework on time.
Next, we plan to track course completion time, satisfaction scores, variability in results and qualitative feedback.
Frequently Asked Questions
What is AI workforce development?
AI workforce development is the structured process of preparing employees to work effectively and responsibly with artificial intelligence. It includes assessing current skills, redesigning roles and workflows, delivering hands-on training and managing change to build the right level of trust in AI tools.
What must remain a human responsibility when using AI in a GxP environment?
AI can change how work is done, but not who is accountable for it. Generative AI outputs must be reviewed and cleared by an authorized person before they are used, and the quality unit remains accountable for each batch.
What skills do employees need to work with AI?
Gartner identifies six capabilities in two areas: work-orchestration skills (problem formulation, systems thinking and strategic foresight) and oversight and governance skills (AI judgment and risk mitigation, human-AI mediation and ethical leadership).
What is AI overreliance?
Overreliance happens when people trust AI outputs without checking them. The first time someone uses AI for a task, they review the output carefully; by the fourth time, they may approve it without reading it. Human-in-the-loop controls only work when the person in the loop is trained to question the output.
AI changes the work, not the accountability
AI is already capable, and it keeps getting better. In regulated manufacturing, what holds it back is not the technology but the operating environment, the regulatory requirements and the people around it. That makes workforce development in pharma, biotech and medtech the foundation of any AI strategy, not a fix after a launch fails.
The goal is not to train people to use more tools. It is to build the judgment to know when an output can be trusted, the habit of checking it and the skills regulators still hold people accountable for. AI changes how work is done, not who is accountable for it. The manufacturers that get value from AI will be the ones that train their people to question it, not just to use it.
Ready to prepare your team to work with AI? Talk to LaSalle’s People & Performance team.


