With decades of experience in management consulting, Marco Gaietti is a seasoned expert in Business Management whose expertise spans strategic management, operations, and customer relations. As organizations grapple with the rapid integration of artificial intelligence, Gaietti observes a widening “confidence gap” that threatens to stall even the most well-funded initiatives. He argues that the next era of productivity will be defined not by the number of software licenses a company holds, but by the psychological safety and trust employees have in their own judgment. In this discussion, we explore the shift from capability to adoption, the necessity of redesigning work before launching training, and why the most critical human skill in an AI-driven world is no longer knowing the answer, but evaluating it. We will examine how a culture that “models the mess” and prioritizes judgment over mere technical fluency can transform AI from a perceived shortcut into a powerful strategic asset.
Many organizations implement AI in specific functions, yet very few achieve advanced capabilities across the entire enterprise. Why does this disconnect persist even when there is genuine commitment from leadership?
The disconnect stems from a fundamental misunderstanding of what successful integration actually looks like. While McKinsey’s 2024 research indicates that 72% of organizations are using AI in at least one business function, the reality is that only a staggering 4% have developed cutting-edge capabilities across their entire operation. This gap exists because leaders often follow a predictable, mechanical arc: they identify an opportunity, allocate a budget, and launch a training program, assuming the rest will follow. However, capability and adoption are two entirely different beasts, and the technical “know-how” does not automatically translate into a transformation of mindset. I have seen this firsthand where people attend every workshop and complete every follow-up session, yet they still walk away asking, “Why are we doing this?” until they see a clear link between the tool and their daily professional identity.
If the problem isn’t necessarily a lack of skills, what is the “confidence gap” you’ve identified, and how does it manifest in the daily lives of employees?
The AI confidence gap is a psychological barrier where employees struggle to trust their own judgment when an algorithm is part of the decision-making loop. SnapLogic’s 2025 research highlights a stark divide: 70% of managers feel very confident with AI, while only 43% of non-managers share that sentiment. This lack of confidence manifests as a fear of being perceived as lazy or incompetent; in fact, 34% of employees worry that using AI will be seen as “cutting corners,” and 27% fear they will be judged outright for relying on it. Instead of a skills gap that can be closed with a simple certification, we are facing a trust deficit where usage might be rising, but the individual’s trust in their own expertise is simultaneously declining. This creates a “shadow AI” culture where people might use the tools in secret but are too afraid to talk about it openly for fear of appearing less valuable.
In the context of IT services and technical expertise, how does AI challenge the traditional professional identity of experienced engineers?
For decades, professional identity in IT has been anchored in the possession of scarce technical expertise—knowing the intricate systems, specific coding languages, and complex architectures that others did not. When AI begins handling code generation, routine support tasks, and knowledge recall with high efficiency, it forces a genuine and often uncomfortable question about where deep expertise now resides. Experienced engineers are navigating a world where their “technical fluency” is no longer the primary differentiator, leading to a sense of professional displacement. The challenge for these senior professionals is to realize that their value is shifting from the ability to generate a solution to the ability to validate it. They need a clear signal from the organization that it is safe to “learn in public” and that their seasoned judgment remains more vital than ever, even if the manual labor of coding is being automated.
You’ve mentioned that “culture comes before curriculum.” Why is work redesign a more effective starting point for AI adoption than immediate training programs?
We have seen this pattern before with cloud migration and Agile methodologies; these shifts stalled not because people couldn’t learn the new systems, but because they didn’t feel safe or understand the purpose of the change. Only 7% of organizations currently provide clear guidelines on how employees should utilize the time that AI saves them, which is a massive oversight. Without work redesign, training is just an abstract exercise, but when we define exactly where AI should augment a decision and where human judgment must remain the final arbiter, the learning becomes purposeful. Employees need to know what to do with the “empty hour” AI creates, whether that is reinvesting in complex problem-solving or deep-focus work. When that foundation is built first, the transformation moves from being a top-down technology mandate to a meaningful people-driven shift.
How can leadership move from being “prepared” for AI change to actually helping their teams adapt on a practical level?
There is a significant disconnect between perception and action among management; while 77% of employees believe their managers are prepared for AI change, only 64% report that those managers are actually helping the team adapt. To close this gap, leaders need to “model the mess,” which means being vulnerable about their own learning curve rather than just acting as experts. When a senior executive stands up and shares a mediocre AI-generated draft, then walks the team through the ten iterations it took to get it right, they provide psychological safety for the entire organization to be a “beginner.” This type of visible, messy practice is far more influential than a polished corporate memo. It transforms the culture into one where experimentation is the norm and failing small is seen as a necessary step toward mastery.
In an era where AI can produce polished outputs in seconds, how should we be reframing the concept of human expertise?
Historically, being an expert meant knowing the answer, but in the AI era, it means knowing whether the AI’s answer is correct, safe, and contextually appropriate. We can look to MIT Sloan’s EPOCH framework, which identifies five human-centric capability groups: empathy, presence, opinion and judgment, creativity and hope, and leadership. Of these, judgment is the most at risk of being undervalued as we become mesmerized by the speed of AI output. We need to place an equal emphasis on evaluating what AI produces as we do on the “prompting” itself. Real expertise now lives in the ability to apply human nuance and ethical consideration to the raw speed of the machine, ensuring that the final result isn’t just fast, but right for the specific human problem we are trying to solve.
Beyond course completions and license counts, what metrics should organizations be using to measure the true health of their AI transformation?
Standard metrics like monthly active users or certification rates are superficial; they don’t tell you if your employees actually trust the tools or their own judgment. Instead, we should look at “experimentation frequency”—are teams testing new, creative use cases, or are they only using AI for the most basic, safe tasks? We also measure the “openness index,” which tracks how comfortably employees discuss their AI usage across different levels of the hierarchy. Finally, the “everyday adoption rate” serves as a north star, looking at the percentage of the workforce that has integrated AI into their daily workflow in small, meaningful ways. Research across 10,600 workers shows that those who receive more than five hours of hands-on training are much more likely to become regular users, at a rate of 79% compared to only 67% for those with less training, which proves that consistency and openness are the keys to a real mindset shift.
What is your forecast for the future of the workforce as we continue to close this AI confidence gap?
I believe we are entering a phase where the competitive advantage of a company will no longer be determined by who has the most sophisticated technology, but by which organization possesses the most confident and psychologically safe workforce. My forecast is that we will see a dramatic shift away from “efficiency-first” AI implementations toward “judgment-first” models, where the primary role of the human is to act as the strategic anchor for automated processes. As the “openness index” improves and more people feel safe enough to innovate out loud, we will see a compounding effect where small wins—like a proposal improved by AI or a faster client turnaround—become the building blocks of a new corporate culture. The winners will be those who treat AI as a partner that requires human expertise to use well, rather than just a shortcut to higher productivity, ultimately resulting in an environment where human capability and machine intelligence work in a genuine, high-trust concert.
