The question every HRD should be ready for
That scenario isn't hypothetical. It happened to an HR director we know, and it's the clearest way we've found to frame what HR needs to bring to AI right now.
A policy and a training programme answer only part of the question. A policy only matters if you can measure how it's used in practice. And most generic training fails: it isn't tied to people's real work, and with AI moving this fast it's out of date almost as soon as it's rolled out.
What the Board actually needs is a view of the workforce the business will require as AI changes the work. That means answers to questions like these:
- Which work will change, and which roles?
- Which skills gain value, and what capability will you need?
- Where will productivity come from?
- What happens to management and organisation design?
- What will you stop recruiting for?
- What will you start building internally?
- What data shows your people are becoming capable with AI, keeping up as it develops, and turning that capability into business results?
Data creates choices. Without it, decisions about people get locked into AI business cases before HR is in the room. So the HR director has to work with the CEO, CFO, technology leaders and business heads from the start, listen to the Board's concerns, and frankly, help educate them.
AI transformation is people transformation. The HR director should help shape it from the outset.
Our co-founder Rob Barnett has seen this from both sides: first as an HR director in FTSE businesses, then as a COO and MD with P&L ownership who kept responsibility for HR. Technology creates possibilities. Whether a business captures them depends on how people work, how managers lead and how the organisation changes.
Four steps for the HR director
If you were being asked that question tomorrow, here's where we'd start. It's a simple loop, and each step builds on the one before.
- 01
Understand. Get clear on what AI is doing to work, roles and expectations in your market.
- 02
Define. Name the problem you're actually trying to fix: your AI capability debt.
- 03
Measure. What gets measured gets managed. Measure capability, not access.
- 04
System. Take a systemic approach to a systemic challenge, and keep repeating it.
Understand what's happening
Start with an honest read of the landscape. Five things stand out:
- AI adoption is inevitable. The question is no longer whether, but how well.
- The technology is accelerating at a pace most businesses can't match.
- It's on everyone's agenda: investors, Boards, the executive team and employees.
- The focus is on productivity, mostly making today's tasks more efficient.
- A number of market leaders are going further, redesigning work around AI.
Productivity is helpful, if you can capture it. But it isn't transformation. Without workforce redesign, the efficiency gains stay scattered across individuals and rarely show up in the business.
AI adoption is inevitable. Capability isn't
Buying licences is easy. Building an organisation that's genuinely good at using AI is not, and almost every system HR runs is affected:
- Workforce planning: skills still matter, but capability, the ability to apply AI in a real role, matters more.
- Jobs: spans, levels and role content are all changing.
- People processes: most were designed for a world in which people did the work. They will need to change.
- Training and ROI: traditional methods won't survive. Capability and business impact will.
- Managers: most are poorly prepared to lead teams where AI does part of the work.
- Reward: frameworks will need to be rethought.
- Trust: it's being challenged, inside and outside the organisation.
HR will need data to help the organisation make real choices. If administrative work falls, should a role take on more customer contact or more complex decisions? If managers can oversee more work, what happens to spans, layers and their own workload?
And what happens to entry-level roles when AI performs some of the tasks through which people have always learned? HR needs to protect the routes through which junior employees develop judgement. Otherwise a short-term efficiency gain could quietly weaken the future leadership pipeline.
Define the problem: AI capability debt
The technology will keep moving faster than your internal systems can adapt. We call the result AI capability debt: the gap between what the technology could deliver for your organisation and what your organisation is capable of doing with it.
This isn't abstract. Capability debt is a real financial drag on performance, and over time we believe the businesses that manage the gap better than their competitors will win. For some, it will be existential.
Part of the problem is that the systems HR controls were all designed for a world in which people did virtually all of the work:
- 01Job architecture
- 02Workforce planning
- 03Skills frameworks and learning
- 04Performance management
- 05Career pathways
- 06Leadership expectations and reward
Some capability debt is inevitable. But the size of the gap can be managed. The challenge is that most businesses don't know how to measure it, and if you can't measure it, you can't manage it or make informed choices.
Measure capability, not usage
Access to AI doesn't tell you whether an organisation is getting good at using it. Capability does: the application of AI to a person's own job, context and workflows. So do results.
So what does a capable person look like? We've concluded there are six dimensions, each assessed at three levels across the organisation.
- 01Understand
Literacy
Do I understand what AI can and can't do?
- 02Protect
Safe and responsible use
Do I understand data, privacy, security, governance and the boundaries of appropriate use?
- 03Think
Framing
Can I identify the problem I'm actually trying to solve?
- 04Instruct
Specification
Can I explain the context, constraints, output and standard I need?
- 05Do
Application
Can I use AI inside real work, not just demonstrate it in a training exercise?
- 06Judge
Evaluation and reflection
Can I look at what AI has produced and decide whether it's actually any good?
That last dimension matters enormously.
Bad AI that's obviously bad is relatively easy to manage. Very good AI that's occasionally wrong is much more dangerous.
Manage capability as a system
How do you build and measure AI capability across an entire workforce? Not with an AI training programme. What's needed is a capability layer that sits between giving people access to AI and actually getting organisational outcomes from it.
It has to be dynamic, because capability has no endpoint. The technology will keep improving, and capability will always lag behind it. The job is to keep the gap as small as possible, with a loop that runs continuously:
- 01
Assess. Where is our capability today, and where are the gaps across the six dimensions?
- 02
Build. What does each role actually need? Prioritise it, because generic learning doesn't land.
- 03
Share. When someone finds an AI workflow that works, capture it, govern it and spread it.
- 04
Measure. Is behaviour changing? Are workflows adopted, time saved and value created?
Then repeat. Assess, build, share, measure, repeat.
Role context is everything in the Build step. A finance manager needs to learn to use AI as a finance manager: in the context of their work, their tools, their governance and their organisation.
Now imagine that six-dimension capability profile not just for one person, but at every level of the organisation, reshaping itself as AI and your people change:
- 01Individual
- 02Function
- 03Business unit
- 04Management population
- 05Organisation
That's workforce data the business has never had. Where are the gaps across a function? Are managers ready to lead? Is capability improving after investment? Add a capability performance management system on top, and that capability starts to compound.
Not a training event. A capability system.
What HR directors should do now
Some form of capability debt is inevitable, and we think accountability for its size will end up mostly in the HR director's lap. Taking the right approach lets you help the leadership team make AI decisions, and help the Board and remuneration committee oversee the consequences for the business and its workforce.
Don't treat AI as another tech project.Its consequences all end up as people choices, so HR belongs in the room before investment decisions lock in assumptions.
Challenge the business case.Push back on any case that assumes savings without explaining how work, roles and management will change.
Start inside HR.Apply the same discipline to a real HR workflow. The experience will sharpen your advice to executive colleagues and the Board.
Bring data, not anecdotes.Measure demonstrated capability across the six dimensions so you can show the Board where you are and whether it's improving.
Redesign the people systems.Job architecture, career paths, performance and reward all assume people do the work. Start rethinking them now.
Protect the pipeline.Keep routes for junior people to build judgement, even as AI absorbs the tasks they used to learn on.
Our challenge to every HR director is simple: help design the AI-enabled workforce now. Don't wait to manage the consequences.
Frequently asked questions
What is AI capability debt?
AI capability debt is the gap between what AI technology could deliver for an organisation and what the organisation is actually capable of doing with it. Because the technology moves faster than people, processes and management systems can adapt, the gap grows unless it is measured and managed.
Why should HR lead AI transformation?
Because the consequences of AI land as people choices: which roles change, what capability is needed, how managers lead, what gets recruited for and how reward works. HR owns the systems that shape those choices, so the HR director needs to help shape AI investment from the outset rather than manage the fallout later.
What is the difference between AI usage and AI capability?
Usage tells you who has access and who logs in. Capability tells you whether people can apply AI well inside their own role, context and workflows, and judge whether the output is any good. Usage is a starting signal; capability is what turns into business results.
Why doesn't generic AI training work?
Generic training is detached from people's real work and goes out of date quickly because the technology moves so fast. People need to learn to use AI in the context of their own role, tools and governance, and keep learning as the technology changes.
How do you measure AI capability across a workforce?
Assess demonstrated skill across six dimensions: literacy, safe and responsible use, framing, specification, application, and evaluation and reflection. Repeat the assessment over time and view the results by individual, function, business unit, management population and organisation.
What happens to entry-level roles as AI takes on more tasks?
Some of the tasks through which junior people traditionally learned will be done by AI. HR needs to deliberately preserve routes through which junior employees develop judgement, or a short-term efficiency gain could weaken the future leadership pipeline.
Where to start
Go back to the four steps and ask yourself honestly where you stand.
Understand. Do you know which work and roles AI is changing in your organisation, and how fast?
Define. Could you describe your AI capability debt to the Board, and where it's widest?
Measure. Do you have demonstrated capability data, by role and team, across all six dimensions?
System. Is capability being assessed, built, shared and measured continuously, or was it a one-off training event?
If you'd struggle with any of these in front of your CEO, that's your first job.
You can do all of this manually, or you can do it with AcademyAI. We measure each person's AI capability across the six dimensions, build role-specific learning into the flow of work, help you capture and spread the workflows that work, and give HR the workforce data to show the Board whether capability is turning into results. Our free diagnostic shows where your organisation stands today.
Start the diagnosticIf your CEO asked tomorrow for your workforce strategy for AI, what would you show them? We'd love to hear how you're answering that question.