Why Training people doesn't Build AI Capability

People, Culture and Sustained Enablement is the lowest-scoring pillar in the entire AI Maturity Index research baseline, at 13%. Lower than strategy. Lower than data. Lower than technology and process. And it’s lowest for a reason that has nothing to do with how hard the problem is to understand, and everything to do with how organisations have chosen to address it.

BCG’s research is direct on this point: the number one barrier to AI value realisation is not technology. It is people and culture. Most organisations have heard some version of this and responded by running a training session. One. Sometimes a half-day workshop. Then they move on, satisfied the box has been ticked.

That response is precisely why this pillar scores lowest. A training event is not enablement. It’s an announcement that enablement was supposed to happen.

The number one barrier to AI value realisation is not technology. It is people and culture.

— BCG

What is AI enablement, and how is it different from AI training?

AI enablement is an ongoing organisational capability that gives people the literacy, support, and psychological safety to use AI effectively in their work — sustained over time, not delivered once and considered complete. AI training is a single event: a workshop, a course, a webinar. Training can be a useful component of enablement. It is not enablement on its own.

The distinction matters because skills decay and tools change. A training session held six months ago, covering tools and use cases that have since evolved, leaves an organisation with a team that once knew something rather than a team that currently knows something. Enablement is the structure that keeps pace with that change: champions, ongoing learning sessions, shared prompt libraries, and a support model that doesn’t end on launch day.

What are the core components of AI enablement?

AI literacy

Not everyone needs to be a data scientist. Everyone does need to know what to delegate to AI, how to prompt well, and how to verify outputs. Literacy is the baseline competency — without it, people either avoid AI tools entirely or use them without understanding their limitations.

Psychological safety to experiment

Mature organisations reward people who try AI and fail fast. Immature organisations reward risk avoidance. Culture is the invisible constraint on every AI initiative — a team that fears being blamed for an AI mistake will quietly avoid using AI at all, regardless of how much training they’ve received.

Sustained enablement

Not a one-day training event. An ongoing support model — champions, office hours, prompt libraries, use-case repositories — that doesn’t stop at launch day. This is the component most organisations skip, because it requires continued investment rather than a single line item.

Change management as a deliverable

Adoption plans are funded and resourced alongside the technology, not treated as an afterthought. The organisation adapts its structure around AI capability — not the other way round, where AI is expected to fit quietly into roles and workflows that never change.

Why does psychological safety matter for AI adoption?

Psychological safety matters for AI adoption because people will not experiment with a new tool in an environment where mistakes are punished, regardless of how skilled they are or how much training they’ve received. The highest maturity level on this dimension in the Index describes a culture that actively rewards AI experimentation — trying, failing fast, and sharing what works is normalised, not penalised.

The lowest maturity level describes active resistance or significant anxiety about AI among a meaningful part of the team. That anxiety is rarely irrational. It often reflects legitimate uncertainty about how AI will change roles, who decides what ‘good use’ looks like, and what happens if an AI-assisted decision goes wrong. Addressing that requires leadership behaviour, not a slide in a training deck.

How should leadership model AI adoption?

Leadership should model AI adoption by visibly using the tools themselves, talking about AI regularly in normal business contexts, and holding the organisation accountable for progress — not by delegating AI entirely to technical teams or individual enthusiasts and hoping it spreads organically.

The maturity ladder on this question is stark. At the lowest level, leadership largely delegates AI to technical teams or individuals. At the highest, leadership actively models AI adoption: using the tools publicly, discussing it regularly, and holding the organisation accountable for measurable progress. Teams notice the gap between what leadership says about AI and what leadership is actually seen doing. The second one is what changes behaviour.

How doES "people and culture" connect to the other four pillars?

"People and culture" is the pillar that determines whether everything invested in the other four pillars actually gets used. A well-redesigned process (Pillar 4) fails in implementation if the team doesn’t trust it or hasn’t been enabled to operate it. A deliberate technology architecture (Pillar 3) sits unused if people default back to old habits because nobody built their confidence in the new tools. Even strong data readiness (Pillar 2) and a clear executive mandate (Pillar 1) don’t compound if the people expected to act on them haven’t been brought along.

This is why people and culture scores lowest across the research, and why it’s the pillar most consistently underinvested relative to its actual leverage. It is also, in some ways, the easiest to start improving — not because it’s simple, but because the first steps (visible leadership participation, psychological safety, an ongoing support structure) don’t require new technology or a redesigned process. They require sustained attention.

Where does your organisation sit on people and culture?

The AI Maturity Index measures this pillar across seven questions: AI literacy across teams, skills development support, the cultural environment around experimentation, change management discipline, leadership modelling, and how the organisation manages the ongoing nature of AI change. Most organisations score lowest here — not because the problem is harder to solve than the others, but because it’s the one most often treated as done after a single training event.

This is the final pillar in the series. Together, the five pillars — strategy, data, technology, process, and people — give a complete picture of where an organisation actually sits on AI maturity, and which gap is constraining the others.