Your board is starting to ask one deceptively simple question: what percentage of our people are AI-fluent? Most organisations cannot answer it with a number that survives a single follow-up. This guide shows how to produce one you can defend.
Short answer: To measure AI fluency across your workforce, define a target fluency level for every role, assess your people against that standard, and report the percentage who meet it. The number only holds up when it has three things behind it: a clear definition of fluency per role, an honest assessment against that definition, and a named owner accountable for closing the gap. Without those, a percentage is a guess with a decimal point.
Why "what percentage of our people are AI-fluent?" is now a board question
Three forces moved this from an HR aspiration to a board-level question in the space of a year.
- Skills are changing faster than most training plans. The World Economic Forum's Future of Jobs Report 2025 estimates that 39% of workers' core skills will change by 2030, and that 59% of the global workforce will need training over the same period. That is not a distant problem. It is a live planning assumption.
- Tools alone are not moving the needle. An MIT study, "The GenAI Divide: State of AI in Business 2025" from its Project NANDA programme, found that roughly 95% of organisations were seeing no measurable return from their generative AI investments. The common thread was not the models. It was how the work and the people around them were set up.
- AI literacy is becoming an explicit expectation, not a nice-to-have. Regulators have started to write it down. The EU AI Act's Article 4, in force since February 2025, asks organisations to ensure a sufficient level of AI literacy among staff who use AI systems. The exact detail is still settling, so confirm the current wording before you rely on it, but the direction is clear.
Put those together and "are our people ready?" is now a risk-and-value question that lands in board packs. Watch-time from a video library is not an answer to it.
What AI fluency actually is
AI fluency is the ability to use AI well and safely in the context of a specific role. A fluent employee knows what AI can and cannot do for their job, uses it to produce real work, checks its output, and knows where the guardrails are.
It is easier to pin down by what it is not:
- It is not tool access. Giving everyone a licence is not the same as everyone being able to use it.
- It is not enthusiasm. A few vocal early adopters do not make a function fluent.
- It is not pilots. A successful pilot proves a use case. It does not prove workforce capability.
Fluency is a standard that a person either meets or does not, for the work they actually do. That is the whole reason it can be counted.
Four signals that look like fluency but are not
When leaders are asked for a number, they usually reach for whichever proxy is nearest. Each one feels like evidence and proves something else.
| What gets measured | What it actually tells you |
|---|---|
| Hours of AI training watched | That content was played, not that anyone can apply it |
| A handful of confident power users | That you have enthusiasts, not a common standard |
| A count of AI pilots running | That use cases exist, not that the workforce is capable |
| A headline percentage with no method behind it | Very little. It will not survive the first follow-up question |
If your current answer rests on one of these, you do not yet have a number. You have a feeling with a figure attached.
A four-level standard you can measure against
You cannot measure fluency without a definition of "fluent." The simplest definition that works across a whole organisation is a ladder of four levels. This is the standard MASSIVUE Academy is built around.
- Associate. Core fluency. Uses AI safely in daily work and spots where it could help.
- Practitioner. Applied execution. Completes real tasks hands-on with AI.
- Specialist. Role and industry depth. Applies AI deeply inside one function, such as HR, finance, or banking.
- Enterprise. Strategic leadership. Owns AI strategy, governance, and business-wide outcomes.
The levels matter because different roles need different targets. A front-line analyst does not need to be an Enterprise-level leader, and a head of function does not need to hand-build prompts all day. You set the target to the job, not to the person's enthusiasm.
| Population | Typical target level |
|---|---|
| All staff in non-technical functions | Associate |
| Team leads and function change agents | Practitioner |
| HR, finance, and operations specialists | Specialist, in their field |
| Executives and senior leaders | Enterprise |
A realistic shape for many organisations is everyone in a function at Associate, a quarter to a third progressing to Practitioner, and the leadership group at Enterprise. Treat those as planning defaults, not a benchmark. Your mix depends on the work your people actually do.
How to produce a number the board will believe
A defensible fluency number passes three tests. Miss any one and it falls over under questioning.
- Definition. Every role has a stated target level. "Fluent" means something specific for that job, written down.
- Assessment. People are measured against that target, not asked to self-rate. A verified assessment or credential is the evidence.
- Accountability. A named owner is responsible for the gap between target and actual, and for closing it.
Here is the method in five steps.
- Pick one function to start. Finance, customer service, HR, anywhere the board is already watching.
- Set the target level for each role in that function, using the ladder above.
- Assess current capability against those targets. A short verified assessment is enough to begin. Do not rely on self-reported confidence, which tends to run high.
- Calculate the percentage who meet their target. That is your fluency number for the function.
- Name an owner and a date for closing the gap, then repeat for the next function.
Worked example, illustrative only and not a client result. A 120-person finance function sets Associate as the target for all staff and Practitioner for its 20 team leads. An assessment finds that 70 of the 120 meet their target. The honest number is 58% fluent in finance, with a named plan to reach 90% within two quarters. That is a sentence a CFO can take to a board. "We think most people are fine" is not.
How to report it to the board
The output of all this is one sentence the board can act on:
"58% of [function] now meets its target AI fluency level, up from 41% last quarter, with the remaining gap owned by [name] and closing by [date]." (Figures illustrative.)
Three things make that sentence credible.
- It is a trend, not a snapshot. One number means little. The direction means everything.
- It is broken down by function, so the board can see where the risk actually sits.
- It names who is accountable, which turns a metric into a plan rather than a status update.
A sensible first milestone is a defensible number for one function inside 90 days, with the next functions scheduled. You are not trying to boil the ocean. You are trying to replace a guess with a method.
Where MASSIVUE Academy fits
The framework above is useful whether or not you ever work with us. Where MASSIVUE Academy helps is turning the standard into evidence you can show.
The Academy is built on the same four-level ladder, across two pillars: AI and Digital Transformation, and Sustainability and Green Finance. Short, applied micro-credentials stack into 13 full certifications, each tied to a level and, where relevant, to a function such as HR, finance, or banking. Every credential is a verified, LinkedIn-shareable digital badge, which is what lets an assessment become a number you can show rather than assert. To be straight about status: credentials are verified and shareable, CPD recognition is in progress, and the courses are not positioned as government-accredited qualifications.
For a whole function or organisation, the Academy supports cohort enrolment, completion reporting, and central billing, so the percentage updates itself as people certify rather than needing a manual recount. If you want to map your own roles to levels, our AI Workforce Transformation service and the free AI maturity assessment are the practical starting points, and you can browse the full Academy catalogue to see how the levels stack.
For wider context, see our related guides on which companies offer AI upskilling programs in Singapore, on measuring enterprise AI ROI, and on building an AI-ready culture.
Get a free one-page AI Fluency Map for one team. Email hello@massivue.com with one function and its headcount, for example "Finance, 120." We will send back a role-to-level map for that team and a realistic starting number you can take to the board. No charge and no pitch.
Prefer to talk it through? Book a 20-minute fluency mapping call through Talk to an Expert.
Frequently asked questions
What does it mean to measure AI fluency?
It means setting a clear standard for what "fluent" looks like in each role, assessing your people against that standard, and reporting the percentage who meet it. Measuring AI fluency is not counting training hours or licences. It is counting who can actually do the work AI now makes possible in their job.
What percentage of our workforce should be AI-fluent?
There is no single right figure, because the target depends on the role. A workable default is 100% of a function at the entry (Associate) level, a quarter to a third at the applied (Practitioner) level, and leaders at the Enterprise level. What matters more than the headline number is that it is defined per role, measured, and trending up.
How is AI fluency different from AI literacy?
AI literacy usually means basic awareness: knowing what AI is and using it safely. AI fluency goes further. It means applying AI to produce real work in a specific role, and knowing its limits. Literacy is the floor. Fluency is the standard you can build a workforce plan and a board number on.
Can you measure AI fluency without a test?
You can start with manager estimates, but it will not hold up to board scrutiny, because self-reported confidence tends to overstate real skill. A short, verified assessment, or a credential earned by passing one, turns a soft estimate into evidence you can defend when someone asks how you got the number.
How long does it take to get a first defensible number?
You can have a credible number for one function within about 90 days: a few weeks to define role targets and run the assessment, then the first wave of people reaching their target. Start with one function the board already cares about, prove the method, then roll it out to the next.
Who should own the AI fluency number?
A named human, usually the CHRO, Chief People Officer, or Head of L&D, with the CEO or board as sponsor. The most common failure is that no one owns the gap between target and actual, so the number is reported but never moves. Accountability is one of the three tests of a defensible figure.
Are MASSIVUE Academy credentials accredited?
MASSIVUE Academy issues verified, LinkedIn-shareable digital credentials, and CPD recognition is in progress. The courses are not positioned as government-accredited qualifications. The value is a consistent, assessable standard you can map roles to and report against, not a regulatory licence.
Sources
- World Economic Forum, Future of Jobs Report 2025 (first-hand): weforum.org. Figures cited: 39% of core skills changing by 2030, 59% of workers needing training by 2030.
- MIT Project NANDA, "The GenAI Divide: State of AI in Business 2025" (July 2025). The ~95% no-measurable-return finding is the report's headline; verified across multiple independent outlets. The report is a preliminary study of a sampled set of organisations, so treat the figure as directional rather than a universal rate.
- EU AI Act, Article 4 (AI literacy), applicable since 2 February 2025 under Article 113(a). Regulatory detail is still being clarified, so confirm the current text with the official source before relying on it.
This article reflects MASSIVUE's practical experience and our reading of published research. It is general guidance, not legal, compliance, or regulatory advice. Figures and regulations change, so verify time-sensitive claims against the primary source before acting on them.