Defining What It Means to Have AI Skills (Part 1)

This is an online copy of our weekly newsletter, written by Constellation's Chief Distiller and Board Advisor Esteban Kolsky, shares curious and interesting insights and data points distilled from enterprise technology to identify what’s notable. If you'd like to get this newsletter in your inbox weekly, Subscribe here.

Today, we are going to begin to cover the issue of talent acquisition for AI needs. There is no checklist of skills or talents to use when hiring, nor are there effective tests to determine whether candidates possess them. As we did during the last paradigm shift, this will start with change agents in the enterprise showing what it takes to succeed, then apply that to external searches.

First, my take.

The next talent question in enterprise AI is becoming clearer.

At a high level, organizations already understand the kinds of people they will need: people who can work with AI, supervise it, question it, redirect it, and redesign processes around it. What remains missing is that workforce at scale. The current talent discussion often stays too close to job counts, while the larger shift is the movement from a workforce that kept fragmented systems functioning through manual intervention to one that can operate, govern, and improve AI-mediated work (wrote about this last week).

Many of the people now under pressure were performing real enterprise work, even if the work existed, because the systems never fit together cleanly. They were the sneaker-net layer of the software ecosystem: reconciling records across applications, maintaining spreadsheets to bridge system gaps, translating one workflow into another, and resolving exceptions that the architecture never anticipated. As agentic AI takes over more of that connective work, enterprises still need much of that human capability, but applied differently.

The value is moving toward judgment, orchestration, exception design, and process redesign.

The labor market data support that direction. BCG estimates that over the next two to three years, 50% to 55% of jobs in the United States will be reshaped by AI, while only 10% to 15% could be fully eliminated over a longer horizon. That suggests accelerated redesign across the workforce. Microsoft found a similar shift in employer behavior: 71% of leaders said they would rather hire a less experienced candidate with AI skills than a more experienced one without them, and 66% said they would not hire someone without AI skills. AI literacy is becoming a baseline expectation rather than a specialist credential.

And yet, it is not well defined or understood – what do we mean by AI Skills? (that sounds like a next blog piece… no?)

I’ve said before that we are living in a paradigm shift, and that comparison to the early personal computer era still holds. Enterprises did not ultimately need large populations of people who merely knew Word or Excel as products. They needed people who understood how work changed when computing became part of the operating environment.

AI is following the same path. The shortage extends beyond elite technical talent. It includes broad operational fluency. The World Economic Forum’s Future of Jobs Report 2025 says AI and big data, networks and cybersecurity, and technological literacy are the three fastest-growing skill areas. It also says that if the global workforce were 100 people, 59 would need training by 2030, while skill gaps are the single biggest barrier to business transformation for 63% of employers. PwC adds another market signal: jobs requiring AI skills are growing even as total job postings have declined, and those jobs carry a substantial wage premium.

Enterprises are facing what is effectively an intractable problem (long-term to solve, with many unknown variables). We know, broadly, what must be built, but it does not exist yet in sufficient volume, and it cannot be produced quickly through current enterprise training models alone. This is a multi-year transition and, in some respects, a generational one. Education systems move slowly. Corporate learning remains too episodic and too tool-specific. The pipeline needed for an AI-native enterprise workforce begins long before hiring and extends well beyond the current boundaries of enterprise training.

That makes the near-term talent search more pragmatic than visionary. Enterprises may spend too much time looking outside for a finished AI workforce that is not yet available in meaningful numbers. A stronger starting point is the people already inside the building, especially those who used to hold the old, spreadsheet-driven, sneaker-net ecosystem together. Many of them understand the exceptions, dependencies, failure points, and informal logic of the process better than anyone else. The task now is to determine which of those people can become supervisors of AI-driven workflows, designers of exception paths, and owners of a more disciplined, yet dynamically orchestrated, operating model.

This also connects back to the SaaSpocalypse argument from last week. Software repricing and talent redesign are parallel signals of the same transition – they are both seeking the new value chain for the enterprise. One reshapes how vendors and enterprises think about software value. The other reshapes how enterprises think about human contribution when connective labor is no longer the default integration layer. Both are early indicators of direction.

Both point to changes that will take longer to resolve than the market currently acknowledges.

Here are some reading resources:

1. Boston Consulting Group says AI will reshape 50% to 55% of U.S. jobs over the next two to three years, while only 10% to 15% are likely to be fully eliminated over a longer horizon, which supports the view that workforce redesign is the larger issue.
2. Microsoft reports that 71% of leaders would rather hire a less experienced candidate with AI skills than a more experienced one without them, and 66% say they would not hire someone without AI skills, indicating that AI literacy is becoming a baseline hiring criterion.
3. The World Economic Forum says that if the global workforce were 100 people, 59 would need training by 2030, and 63% of employers identify skill gaps as a major barrier to business transformation, which supports the argument that this is a multi-year workforce transition.
4. The World Economic Forum also shows that AI and big data, technological literacy, and networks and cybersecurity are among the fastest-growing skill areas, which help define the broad profile of the workforce enterprises are trying to build.
5. PwC says jobs requiring AI skills rose 7.5% even as total job postings fell 11.3%, which indicates employers are still prioritizing AI-capable talent in a softer hiring market.
6. PwC also says workers with AI skills earn an average wage premium of 56%, reinforcing the point that enterprises are attaching greater economic value to AI fluency.
7. Accenture argues that leading companies are moving from job-based talent models toward task-, skill-, and value-based workforce design, which aligns with the shift from manual bridge work to redesigned human-plus-AI operating models.
8. Microsoft says managers are already planning for new roles, such as AI agent specialists and AI workforce managers, which is useful evidence that the talent model is changing alongside the operating model.
9. IBM argues that agentic AI requires workforce evolution, trust, and an operating-model redesign rather than simple automation adoption, thereby extending the linchpin discussion into governance and execution.
10. Accenture outlines how learning systems need to be redesigned for human-AI collaboration at scale, which supports the argument that this transition is multi-year and cannot be solved by episodic training alone.

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.