AI Skills, Part 1: Experience over Expertise

Welcome to a new edition of The Board: Distillation Aftershots ^(*).

This is an online archive for the newsletter, written by Constellation's Chief Distiller and Board Advisor Esteban Kolsky, that shares curious and interesting insights and data points distilled from enterprise technology to identify what’s notable. Would you like to get this newsletter? Subscribe.

In this issue, I want to move the talent discussion forward. The previous piece focused on the human linchpin and the kind of connective work AI is beginning to absorb. This one focuses on what follows from that: how to think about the people enterprises will actually need when AI becomes part of the operating model rather than an overlay on top of it.

First, my take.

The market still talks about AI talent as if it were a shopping list of talent that we are seeking. Alas, enterprises should not be looking for a clean inventory of discrete AI skills nearly as much as they should be looking for people who can work in an environment where ambiguity is constant, context is incomplete, and the first answer is rarely the right one.

The more useful people in this next cycle will be those who can take a vague problem, give it shape, decide which part belongs to a machine, and then stay with the output long enough to see whether it survives contact with the actual business process. That is why I keep coming back to experience over expertise. Expertise is often too narrow for what comes next; it is, by now, commoditized. Enterprise AI work sits in the gray space between functions, systems, and decisions. It needs people who can move through that ambiguity, recognize what is missing from the question, and keep refining until the result becomes usable.

This is also why I do not think the answer lies in chasing ready-made AI talent as if there were enough of it to buy. There is not. Microsoft reported in 2024 that 66% of leaders would not hire someone without AI skills, and 71% would prefer a less experienced candidate with AI skills over a more experienced one without them. By 2025, Microsoft was also reporting that 83% of leaders believed AI would let employees take on more complex and strategic work earlier in their careers. Those numbers are evidence that the market already senses the shift, even if it still describes it badly.

The enterprises that handle this transition best will likely favor people who can structure the unstructured, see beyond the obvious answer, and understand that cognition is iterative. Good work with AI usually comes from framing, testing, revising, redirecting, and repeating. That is less a tool skill than a way of thinking. It does not need people who can automate a process; it needs people who can create a new way to do business, using AI as a resource.

This is where the blend of scientific and humanities habits becomes more useful than most enterprise hiring language allows. The scientific side helps with structure, causality, disciplined experimentation, and the habit of testing assumptions. The humanities side helps with language, interpretation, context, meaning, and understanding what a human is actually trying to accomplish. The World Economic Forum’s Future of Jobs Report 2025 puts creative thinking alongside AI and big data, technological literacy, and networks and cybersecurity among the fastest-growing skill areas. The next workforce will need both technical fluency and judgment.

We know, at a high level, which human contributions will matter more in an AI-mediated enterprise. Building that at scale is another matter. The World Economic Forum says 59 out of every 100 workers worldwide will need training by 2030, and 63% of employers identify skill gaps as the main barrier to business transformation. Accenture adds a useful operational point: only 11% of organizations are equipped to enable effective co-learning between humans and AI. That is not a short-cycle training issue. It is a longer redesign of how people learn, adapt, and work.

That is where the current conversation still falls short. Enterprises are asking for finished talent profiles when they really need people who can think clearly in shifting conditions, impose order on messy environments, and work with machines without assuming the machines understand more than they do. Humans still tell computers what to do. Computers still do as they’re told. The space between those two facts is where much of the next talent value will sit.

Here are some reading resources:

1. OECD examines the AI skills gap from the supply side of training and shows that current course availability remains too limited to meet the growing demand for both general AI literacy and advanced AI capability.
2. OECD examines what it takes to build an AI-ready workforce within institutions, with an emphasis on internal capabilities, leadership, governance, and the operating conditions required to use AI effectively.
3. The OECD Skills Outlook 2025 provides a broader skills and capabilities framework that supports the argument that AI readiness is part of a longer-cycle adaptation in learning, work, and problem-solving.
4. UNESCO reports on how thousands of companies are approaching AI training and governance, which is useful for showing that many organizations recognize the need but still lack mature and comprehensive preparation.
5. UNESCO-UNEVOC provides a youth-focused view of digital and AI readiness that supports the generational dimension of the workforce transition and the longer-term pipeline challenge.
6. UNESCO’s AI in education work is relevant because it frames the transition as an education-system issue as much as an enterprise-training issue, which aligns with the broader argument about long-cycle talent formation.
7. Accenture’s Talent Reinventors report explains how work is being broken down into tasks, skills, costs, and value in a human-plus-AI environment, which aligns with the shift away from static roles and toward redesigned contributions.

What’s your take? We are fostering a community of executives who want to discuss these issues in depth. This newsletter is but a part of it. We welcome your feedback and look forward to engaging in these conversations.

If you are interested in exploring the full report, discussing the Board’s offering further, or have any additional questions, please contact me at [email protected], and I will be happy to connect with you.

(*) A normal distillation process produces byproducts: primary, simple ones called foreshots, and secondary, more complex and nuanced ones called aftershots. This newsletter highlights remnants from the distillation process, the “cutting room floor” elements, and shares insights to complement the monthly report.