AI literacy is becoming organisational infrastructure
- trinimaturana
- Tendencias y Futuro, The Voices in English
Índice
There is a moment most organisations have probably already lived through, even if they have not named it yet.
A team completes an AI training session. People leave with new skills, a better sense of what the tools can do, maybe even some excitement about the possibilities. A few weeks later, someone includes an AI-generated output in a client presentation without checking it properly. Someone else uploads a sensitive internal document to a public tool that was never on the approved list. A manager approves a recommendation shaped, invisibly, by a hallucinated data point.
Nobody acted in bad faith. Nobody ignored the guidance. Everyone completed the training.
And yet something failed that the training was never designed to prevent.
That is the gap many organisations are still struggling to see clearly. Not the gap between having AI and using it. The gap between using AI and thinking well in its presence.
The assumption that deserves more scrutiny
The pressure to act on AI literacy is understandable. Regulators are moving. Leadership teams are under pressure to show progress. AI adoption is accelerating faster than governance structures can stabilise. Under the EU AI Act, organisations deploying certain AI systems are now expected to ensure an adequate level of literacy among staff and relevant stakeholders.
So the response has been predictable. Organisations train. They launch workshops, create internal playbooks, circulate prompt guides and roll out awareness campaigns explaining what responsible AI use should look like.
The underlying assumption feels perfectly reasonable: if people know how to use AI, the organisation becomes AI-ready.
But that assumption deserves more scrutiny than it is currently getting.
Because using AI and operating intelligently in an AI environment are not the same thing.
McKinsey’s 2025 research exposed that contradiction in a way that is difficult to ignore. While 92% of companies plan to increase their AI investments over the next three years, only 1% describe themselves as mature in how they actually deploy it.
That gap does not exist because organisations lack access to tools.
It exists because access is scaling faster than organisational intelligence.
The issue is not whether employees can generate outputs. The issue is whether the organisation knows how to think with AI.
Literacy without judgement creates false confidence
And that becomes even more uncomfortable when you look at behaviour.
KPMG’s 2025 global study on trust, attitudes and AI use found that 66% of employees using AI rely on outputs without evaluating their accuracy.
That number says something important.
Not because employees are careless. Because it reveals a version of literacy that remains dangerously incomplete.
These are not people struggling with prompts or interfaces. They already know how to use the tools. What they have not necessarily developed is judgement.
Generating an answer is easy. Knowing when that answer deserves scepticism is something else entirely.
Speed feels productive. Recognising when speed introduces risk requires a different kind of maturity.
Uploading information takes seconds. Understanding the implications of where that information goes requires awareness that many organisations have barely begun to build.
This is where many AI literacy conversations still stop too early. They focus on usage because usage is visible and measurable. But capability lives somewhere deeper.
In any complex organisation, judgement is never purely individual. It lives in shared norms, in leadership behaviour, in the quality of conversations teams have with one another, and in the questions organisations teach people to ask.
That changes the conversation entirely.
The questions training cannot answer
Because what employees actually need in an AI-saturated environment is not simply technical familiarity.
They need to know how much trust to place in AI-generated outputs depending on the context. They need clarity about which decisions require human escalation and why. They need a shared sense of what constitutes acceptable uncertainty before action is taken. They need to understand how authenticity should be evaluated when content generation becomes frictionless.
None of those are technical questions.
They are interpretive ones.
And interpretation cannot be installed through a workshop.
One of the more thoughtful European contributions to this conversation, the ARISA AI Literacy Practices Report, makes that distinction clearly by framing AI literacy as something that extends beyond understanding how systems function to include critical thinking, ethical awareness and informed decision-making.
That distinction matters enormously.
Because organisations that interpret literacy as technical familiarity are preparing employees to use tools.
Organisations that interpret literacy as collective judgement are preparing themselves to operate differently.
Those are profoundly different ambitions.
Meaning is where organisational capability gets built
This is also where internal communication enters the conversation in a way many organisations are still underestimating.
AI literacy is usually assigned to IT, learning teams or compliance functions. That makes sense on the surface. Those teams own the systems, the governance structures and the formal learning mechanisms.
But there is something they do not typically own.
Meaning.
AI adoption is not simply a technology rollout. It changes how people solve problems, how they validate information, how they build confidence in decisions, and how they interpret authority. Those are not implementation questions. They are cultural ones.
And culture does not move because a policy document exists.
It moves through repeated organisational sensemaking. Through conversations that help people interpret uncertainty. Through shared language that allows teams to reason together when something new starts reshaping how work gets done.
That is precisely where internal communication becomes strategically relevant.
Not as the team teaching prompting.
Not as the function translating the AI strategy into cleaner messaging.
As the function helping the organisation build collective judgement.
That means making space for tensions leadership may not yet be seeing. The team quietly uncomfortable with how much their manager relies on AI-generated analysis. The communicator wondering whether content produced with AI still reflects the organisation’s actual voice. The leader approving recommendations without fully understanding the reasoning behind them.
These are not unusual exceptions.
They are the texture of organisational life right now.
And they will not surface in governance frameworks.
They surface in conversation.
Internal communication is uniquely positioned to make those conversations possible.
That may be one of the most strategic mandates the function has inherited in years.
What mature AI literacy actually looks like
The organisations that scale AI intelligently will not necessarily be the ones with the most sophisticated platforms or the highest training completion rates.
They will be the ones that develop something harder to measure and far more valuable: a collective instinct for when to trust, when to question, when to slow down, and when human interpretation remains non-negotiable.
That capability does not emerge from a workshop.
It emerges from the quality of the conversations an organisation sustains over time. From leadership willing to model uncertainty instead of projecting artificial certainty. From communicators helping teams develop shared language around credibility, judgement and risk.
AI literacy, understood this way, stops being a learning objective.
It becomes organisational infrastructure.
Organisations are investing heavily in teaching people how to use AI. Far fewer are building the collective judgement required to live with it.
Signals shaping this reflection
McKinsey & Company (2025)
Superagency in the workplace: Empowering people to unlock AI’s full potential at work
KPMG & University of Melbourne (2025)
Trust, attitudes and use of Artificial Intelligence: A global study
ARISA / AI Skills Alliance (2025)
AI Literacy Practices Report
European Union AI Act
Article 4 and organisational AI literacy obligations