AI and the Workforce: Answer the Board by Role, Not by Sentiment

The Board Will Ask About AI and Your Workforce. Will You Have Data or a Point of View?

The Board is asking about AI, can you answer by role? 

Somewhere in the next two Board cycles, a director is going to ask what AI means for your workforce. Most CHROs can answer that question with a point of view. Very few can answer it with data. 

The point-of-view answer usually sounds like this: we’re taking it seriously, we’re piloting tools, we’re investing in reskilling. That’s sincere and it’s true, but it answers a feelings question, not an operating one. What the Board actually wants to know is which roles are exposed, by how much, and what it costs you to wait. 

The sentiment answer fails quietly 

The problem with answering on sentiment isn’t at the Board meeting. It’s what happens after. Organizations that commit reskilling and redeployment budgets without a role-level picture of exposure end up making broad bets, enterprise-wide learning programs, blanket upskilling initiatives, that spread investment across people who didn’t need it and miss the ones who did. 

The challenge for CHROs is coming up with a dynamic talent strategy that lines up with AI agents’ capability to run the business more efficiently. This includes knowing what skills they currently have, how to retain them, and what skills they need to acquire. They also need to balance the employee experience, and the corporate messaging around the change that is happening.   

Targeted works better than broad, but only if you know where to aim. That means building a real data product: AI exposure scored by role, job family, and function, using validated labor-market intelligence layered against your own workforce data. 

This sounds overwhelming, but what I tell my clients is to bring it back to simplicity. Whether you’re designing reports, turning on dashboards, or creating targeted training packages, it all comes back to the quality of your data. Considerations for turning on HR AI agents are no different and should start with good hygiene at the data level. Are your organizations and job structures set up efficiently? Are skills for jobs uniform, meaning are they identified the same way and do they mean the same thing across your organizations and locations? By making sure your data is clean and uniform across your organizations, you increase the efficiency of the AI agents you invoke and your company’s ability to compete. Our solution to this is much simpler than you may think.  

Four weeks of work, not a transformation program 

The instinct when the Board asks a big question is to launch a big initiative. Resist it. The right first move is deliberately small: one functional group, a contained set of job profiles, real data, four weeks. The output is a prioritized, role-level risk picture your leadership team can act on, and a defensible basis for every reskilling dollar that follows. 

A pilot that produces signal in four weeks is worth more than a program that produces slides in twelve months. If the risk picture warrants going wider, you widen with evidence. 

Where to start depends on where you are 

Stabilize. If your workforce data isn’t governed, if roles are inconsistent or job architecture has drifted, any AI risk picture you build on it won’t hold up. Fix the foundation first. 

Optimize. If the data is sound, produce the map: role-level exposure for a pilot scope, and let the evidence set the agenda. 

Modernize. The map isn’t the end state. Fed into reskilling, redeployment, and workforce planning, it becomes a running capability: the workforce answer to the AI question, maintained continuously instead of assembled once a year. 

The Board-meeting test 

One question tells you where you stand: if the Board asked tomorrow which roles in your organization are most exposed to AI, could you answer by role, or only by sentiment? 

If the honest answer is sentiment, the gap is closeable, and it’s cheaper to close on your schedule than on the Board’s.