The talent market cannot supply AI governance capability at the volume organisations now require. Across the Asia-Pacific, only 21% of organisations believe they can recruit or retain people with this capability (Aon, 2026). In the same region, nearly three-quarters have already deployed or piloted AI (Aon, 2026). What the market offers is compliance expertise and risk management knowledge: people who know which assessments a framework requires. That is not the same as knowing what to do when the assessment comes back ambiguous and the system is already live.
In Brief
- Governance judgment is built through structured developmental exposure — it is not a credential the talent market can recruit or supply.
- Most AI adoption programs were designed for deployment speed and produced no conditions under which governance judgment could develop.
- Hiring for AI governance roles addresses a vacancy; it does not resolve the underlying developmental deficit.
- Any active AI program can be redesigned to create the conditions under which governance judgment develops.
- The difference between organisations that are building governance capability and those that are not is program design intent.
Governance judgment — the capacity to make defensible decisions about AI use in genuinely novel situations, where the framework cannot specify the right answer — is built through structured developmental exposure. It requires repeated engagement with consequential decisions, observation of outcomes, and accumulated understanding of where AI systems diverge from what governance frameworks assumed they would do. That developmental exposure was not designed into most organisations’ AI adoption programs. The programs were built for deployment, not for the capability development that governance judgment requires. The market reflects exactly that design decision.
The market never built governance judgment
Governance judgment is produced by specific organisational conditions, not by career history in AI.
A career in AI deployment does not produce AI governance judgment. The technical expert who has spent five years deploying AI systems has accumulated deep understanding of how AI works, where it performs reliably, and what can go wrong technically. What this does not produce is the judgment required to decide, in real time, when AI output is reliable enough to act on in a high-stakes context, when a governance boundary has been reached, and when escalation is warranted rather than a technical fix. These are not problems that yield to expertise in AI itself. They yield to structured exposure to governance decisions under conditions where the outcomes are visible, the feedback is timely, and the stakes are real.
AI adoption programs were built for deployment speed; capability development was not part of the design intent. Timelines were set around deployment milestones: pilots launched, models trained, systems integrated. The governance activity that ran alongside deployment was largely framework compliance — completing assessments, documenting risks, establishing policies — and all of it necessary. But it does not produce governance judgment, because it does not require decision-making in conditions of genuine ambiguity. Governance frameworks specify what to assess. Governance judgment is exercised when the assessment does not specify what to do next.
Organisations built their AI governance thinking by importing frameworks from adjacent disciplines: technology risk, information security, regulatory compliance. These disciplines are designed for knowable problems — environments where the right answer exists and the governance function is to ensure it is followed. AI governance requires something different: judgment about genuinely novel situations where the technology behaves in ways the framework did not anticipate, and where the decision must be made before the evidence is complete. No prior discipline has built the developmental infrastructure for this capability at scale. The market has nothing to draw on because organisations have not yet created the conditions under which it could develop.
The structural condition is not unique to AI. Clinical expertise does not produce hospital governance judgment; engineering mastery does not produce safety culture at system level. In each domain, governance judgment required deliberate institutional investment in how decisions were structured — not simply in who held the roles.
The market has nothing to draw on because organisations have not yet created the conditions under which it could develop.
Recruitment does not build governance judgment
Going to market for governance talent fixes the vacancy but not the deficit.
When an organisation hires a Chief AI Officer or a Head of AI Governance, it acquires a person with strong credentials in adjacent disciplines — technology risk, data governance, policy frameworks — who will develop governance judgment on the job, in the organisation’s specific context, over time. The underlying condition remains: governance judgment for AI programs develops inside each organisation, in context, over time. It is not a portable credential that a hiring process can supply.
The consequence of treating this as a recruitment problem is that organisations run several cycles of AI program delivery without the governance capability their programs require, and without building it. Each program completes, governance arrangements are reviewed, frameworks are refined, and the next program begins. The team that governed the previous program has accumulated experience, but that experience rarely survives a restructure or staff rotation in any structured or transferable form. The organisation ends each cycle at roughly the same governance capability it started with. The cumulative learning that should have built the capability did not accumulate in a form the organisation can use. As AI programs scale and the decisions they require become more consequential, this capability deficit becomes more exposed; no hiring decision reaches the conditions that produced it.
Programs create what markets cannot supply
The developmental conditions governance judgment requires can be structured into any active AI program; the question is whether it was designed with that intent.
This does not require a separate investment. It requires a different design intent applied to the programs already running. Governance judgment develops when people are making decisions with real consequences that are reviewed after outcomes are known; not when they are processing assessments at defined checkpoints. When governance sits at the intersection of technical and business judgment, rather than being separated into a compliance function that reviews output after delivery, the judgment required is genuinely exercised. Organisations whose governance people are making consequential decisions at that intersection are building the capability. Those whose governance people are completing compliance documentation are not. The difference is program design, not the people in the roles.
The executive who runs an AI program with developmental intent is not adding overhead. They are producing governance capability as a consequence of the work itself, rather than expecting a talent market to supply what it was never asked to build. That is a different program design, and a materially different governance position for the organisation across successive cycles.
What this means for senior leaders
- AI governance capability cannot be recruited from the talent market — it develops inside AI programs through structured exposure to consequential decisions, or it does not develop at all.
- The test for readiness is not whether governance roles are filled; it is whether your AI programs are creating the conditions under which governance judgment develops.
- Examine whether your governance people are making consequential decisions with review cycles, or completing assessments at defined milestones. The former builds the capability; the latter processes paperwork.
- Governance judgment is a product of how AI programs are designed, not a property of who is hired into them. The executive who understands this is building a capability their peers will still be trying to recruit in three years.
References
- Aon plc. (2026). Human Capital Trends 2026 Study. Aon. https://www.aon.com/apac/in-the-press/asia-newsroom/2026/apac-leads-in-ai-adoption-but-lags-in-workforce-readiness-aon-study-helps
