There Is No Universal AI-Ready Workforce: Why Every Organisation Needs a Different AI Skills Mix

Evidence from government, academia and professional competency frameworks suggests that AI readiness should be defined not by how much AI employees know, but by whether they possess the right capabilities for their responsibilities.

The problem with treating AI literacy as a universal skill

As artificial intelligence becomes embedded in everyday business operations, organisations are investing heavily in developing the AI capabilities of their workforces.

Much of this investment is built around a relatively simple assumption: employees need to become more proficient in AI.

Training programs introduce generative AI, prompt engineering, automation, responsible AI and the practical application of AI tools. Employees are encouraged to improve their AI literacy, while organisations attempt to measure their progress toward becoming more AI-ready.

But this approach raises a fundamental question. Should every employee be developing the same AI competencies?

Consider three employees working within the same organisation.

A financial analyst uses AI to interrogate forecasts, investigate financial performance and evaluate scenarios. Their competence depends on understanding how to work with AI-generated analysis, assess the reliability of outputs and exercise appropriate financial judgement.

A human resources manager may be responsible for overseeing AI-assisted recruitment, workforce planning or employee assessment. Their competency requirements extend into algorithmic bias, privacy, fairness, human oversight and the potential consequences of automated decision-making.

Meanwhile, a chief executive may rarely interact with these systems directly but is responsible for determining where the organisation should invest in AI, which processes should change, how risks should be managed and how AI will influence the company's future operating model.

These individuals require some common foundational knowledge, but the AI competencies most relevant to their responsibilities are fundamentally different.

This distinction leads to an important conclusion. If the AI competencies required of individuals depend on their roles, responsibilities and influence, then the optimal distribution of AI competencies across an organisation must depend on the composition and ambitions of that organisation.

Increasingly, government frameworks, professional standards and academic research support this proposition.

AI competency is already being defined by responsibility, not just technical proficiency

In November 2025, the UK's Skills England published its AI Skills Tools Package, developed using evidence from six national workshops, an expert roundtable held at the British Academy and supporting desk research.

The package's AI Skills Framework maps AI skills across three job levels: entry level, mid level and managerial.

At entry level, the emphasis is on capabilities such as interacting with AI tools, evaluating outputs, understanding limitations and protecting information.

At mid level, the emphasis expands toward embedding AI into workflows, coordinating adoption and exercising professional judgement over AI-supported work.

At managerial level, it introduces competencies associated with organisational AI strategy, investment, governance, accountability and managing AI-enabled transformation: the decisions that shape how AI is adopted across the organisation.

Importantly, the framework states that its levels are not tied to specific sectors or job titles, and that individuals may span more than one level depending on their role. It also describes itself as a starting point, noting that technical and ethical demands vary significantly between contexts.

The significance is clear. The UK government is not simply recommending that senior employees receive more advanced versions of the same AI training. It is recognising that different organisational responsibilities demand different combinations of capabilities.

The Alan Turing Institute reaches a similar conclusion through its AI Skills for Business Competency Framework, developed with support from the UK's Department for Science, Innovation and Technology and Innovate UK BridgeAI.

Rather than defining a universal AI user, the framework distinguishes between AI Citizens, AI Workers, AI Professionals and AI Leaders.

AI Workers are employees whose primary occupation is not AI but whose performance can be enhanced through it. AI Professionals have core responsibilities involving data and AI. AI Leaders carry responsibility for matters such as AI adoption, organisational strategy and governance.

This differentiation matters because the framework recognises that competence is not limited to technical proficiency. Leadership, evaluation, responsible adoption and organisational decision-making are themselves AI-related competencies.

A consultation draft of Version 3 of the framework, published in December 2025, substantially updates its competency domains, reinforcing the importance of contextualising competency requirements.

Together, these frameworks establish a common principle: AI competence should be defined in relation to what an individual is responsible for accomplishing, rather than through a generic measure of AI knowledge.

The EU AI Act reinforces the importance of context

This principle is also reflected in European regulation.

Article 4 of the EU AI Act places an AI literacy obligation on organisations that provide or deploy AI systems, covering their staff and others operating AI systems on their behalf. From the outset, the article has required those measures to take account of people's technical knowledge, experience, education and training, the context in which the AI systems are used, and the people on whom they are used.

In July 2026, the obligation was amended by Regulation (EU) 2026/1744, part of the EU's Digital Omnibus package. Providers and deployers no longer have to ensure a sufficient level of AI literacy. They must take measures to support its development, and the amended text states that this does not require them to guarantee any specific level of AI literacy for any individual.

The European Commission's guidance on Article 4 is explicit that a single template will not do:

"As there is no one-size-fits-all approach for AI literacy, there is also not one unique framework."

The same guidance notes that no strict requirements or mandatory trainings are imposed, and that appropriate measures depend on the AI systems involved, their purpose, their risks and the people interacting with them.

This is an important regulatory distinction. Article 4 does not prescribe a universal level of proficiency or require organisations to guarantee that every employee reaches a particular competency standard. It places responsibility on providers and deployers to support literacy in ways appropriate to their circumstances.

A complementary policy brief published by the European think tank Interface in August 2026 goes further. Its Operational Framework for AI Literacy in the Workplace proposes three cumulative tiers, Baseline, Operational and Oversight, assigned according to the nature and consequences of a person's interactions with AI.

Its central argument is that job titles and seniority are the wrong basis for setting literacy requirements. In the authors' words: "An individual's tier follows their role, not their seniority or technical background." Their own illustration contrasts an accountant using AI in their work with an operator overseeing a credit-scoring model, each of whom needs different knowledge and judgement.

The same logic extends across any organisation. Someone overseeing an AI system that influences hiring decisions needs substantially different competencies from someone using AI to draft marketing materials, even when both sit at a similar level in the organisational hierarchy.

The implication is that AI competence is multidimensional. It must account for the tasks people perform, the decisions they influence and the consequences associated with those decisions.

Organisational influence is an important determinant of required competence

This argument is further supported by the Skills Framework for the Information Age, commonly known as SFIA.

SFIA provides an established structure for understanding professional competency through seven levels of responsibility, described using generic attributes: autonomy, influence, complexity, knowledge, and business skills and behavioural factors.

Applied to AI, this creates a useful distinction between the skills required to perform a task, the skills required to supervise its execution, and the skills required to decide whether and how that task should be delegated to an AI system.

A professional using AI to complete an operational activity may need strong capabilities in application, verification and judgement.

A manager responsible for introducing AI into that activity may additionally require competencies in workflow redesign, performance measurement, change management and human oversight.

An executive authorising AI adoption may instead require expertise in strategic prioritisation, investment evaluation, organisational risk and accountability.

These are not necessarily progressive levels of the same capability. They are different competency requirements arising from different responsibilities.

SFIA's own framework for AI skills, first released alongside SFIA 9 in October 2024, places AI skills within this existing responsibility structure rather than treating them as a separate specialism, and encourages organisations to combine it with methods and standards that match their own AI implementation needs.

This reinforces an important proposition: organisational influence may be just as relevant as technical proficiency when determining which AI competencies an individual should possess.

Academic research supports differentiated AI literacy

The distinction between organisational roles and AI capabilities is also emerging in academic research.

In October 2025, Alexander Benlian and Marc Pinski published The AI Literacy Development Canvas: Assessing and Building AI Literacy in Organizations in the journal Business Horizons.

Their research introduces an AI Literacy Assessment Matrix that relates AI literacy expectations to organisational level, rather than assuming that every employee requires the same degree of proficiency.

The research is particularly relevant because it explicitly separates an organisation's current AI capability from the capability it needs to develop. Its development canvas is a planning tool for deciding where and how to invest in AI training: assess existing literacy, define the target, and design interventions to close the resulting gap.

The paper draws on academic literature and managerial insight, with an illustrative pharmaceutical-sector case study. It therefore offers a structured conceptual and practical approach, rather than experimental proof of the effectiveness of every proposed intervention.

Nevertheless, its central proposition closely supports the case for role-sensitive AI competency development.

Real-world AI usage also differs across organisational functions

Beyond conceptual frameworks, emerging empirical evidence suggests that AI proficiency and usage patterns already vary significantly across business functions.

A working paper released in August 2026 by Nicholas Hallman, Zachary Kowaleski, Anu Puvvada and Jaime Schmidt, titled Sophistication in GenAI Use: Field Evidence from a Large Firm, examined 713,564 employee prompts and the corresponding AI responses at a single large firm, which the paper does not name.

The study covered nearly 4,000 back-office employees across 15 functional areas over eight months in 2025.

Its findings showed that more senior employees tended to exhibit more sophisticated generative AI use. The researchers suggest that domain expertise may complement the ability to use AI effectively.

They also identified substantial differences between functions, with Strategy, Digital Innovation and Project Management exhibiting the highest levels of sophistication.

Notably, the researchers observed neither improvements in sophistication over time nor lasting improvements following formal AI training within the study period.

These findings are particularly relevant to organisations attempting to build AI-ready workforces. They suggest that AI proficiency is shaped by professional context, and that simply providing standardised AI training may not be sufficient to produce sustained improvements in capability.

The study does not prove which competencies employees should possess. It measures observed patterns within one firm, and it gauges sophistication from how prompts are written and used rather than from the quality of the resulting work, so its findings should be read accordingly.

However, it does provide direct evidence that actual AI capability is distributed unevenly across functions and levels of seniority.

Further research suggests that AI adoption is changing the nature of managerial competence itself.

A 2026 study in the Strategic Management Journal, Artificial Intelligence Adoption and the Demand for Managerial Expertise, examined the relationship between organisational AI adoption and demand for managerial skills using job-posting data.

The researchers found that greater AI adoption was associated with increased demand for managers, and with a shift toward interpersonal and growth-oriented skills such as stakeholder management, creativity and sales management. Demand moved away from more traditional managerial capabilities, including budgeting and staff management.

The study identifies associations rather than establishing that AI directly caused every change. Nevertheless, it reinforces the idea that as organisations adopt AI, the skills required to perform existing roles change.

From individual competencies to organisational capability

Taken together, these findings support a broader conclusion. If AI competency requirements vary according to individual responsibilities, it follows that an organisation's required mix of AI capabilities depends on how those responsibilities are distributed across its workforce.

Consider a professional services firm employing thousands of consultants, analysts, auditors and client relationship managers. Its desired AI competency profile might emphasise analytical reasoning, evaluation of AI-generated outputs, research, professional judgement, client confidentiality and the responsible integration of AI into advisory work.

A manufacturing organisation may have a different profile, reflecting the importance of operational automation, quality control, supply chain optimisation, predictive systems and human oversight of AI-enabled processes.

A financial institution may require comparatively greater emphasis on model risk, governance, regulatory compliance, auditability and decision accountability.

Each organisation still needs a common foundation of AI literacy, but the relative importance, breadth and depth of additional competencies will differ. Even two companies in the same industry may need different competency profiles if their operating models, workforce structures, technologies or AI strategies differ.

This does not mean every organisation must possess a mathematically unique distribution of competencies. Organisations with similar characteristics may have similar requirements. Rather, there is no reason to assume that a single, universal distribution of AI competencies would be optimal for every organisation.

This is a significant departure from conventional approaches to workforce AI readiness. Instead of asking whether employees have achieved a common standard of AI proficiency, organisations should ask whether their workforces collectively possess the competencies needed to execute their particular strategies and responsibilities.

The critical distinction between actual and desired AI capability

This introduces two distinct organisational measures.

Actual AI capability describes the competencies employees currently possess. It reflects existing knowledge, practical skills, professional experience and the ability to use or oversee AI within the organisation.

Desired AI capability describes the competencies employees should possess to perform their responsibilities effectively in the organisation's intended AI-enabled operating model. It reflects the demands of their roles, their degree of organisational influence, their decision-making authority, the risks associated with their responsibilities and the strategic objectives of the organisation.

The difference between these two states is the organisation's AI capability gap. Importantly, that gap cannot be fully understood through a single average AI literacy score.

An organisation could perform well on a generic assessment of AI knowledge while lacking critical capabilities in AI governance, workflow transformation or professional oversight. Equally, an organisation might have exceptionally capable AI practitioners but lack the leadership expertise required to identify valuable opportunities, approve investments or manage organisational change.

The distribution of competency matters as much as the overall level.

This perspective is consistent with Benlian and Pinski's recommendation to distinguish existing AI literacy from target competencies, and with the UK government's emphasis on identifying skill gaps relative to organisational adoption needs.

It also changes the purpose of AI assessment. Rather than measuring AI knowledge in isolation, assessment becomes a means of determining whether individuals and organisations possess the capabilities required of them.

The optimal AI capability profile will continue to evolve

There is another consequence of this argument. An organisation's desired distribution of AI competencies is unlikely to remain fixed.

As AI capabilities improve, previously manual workflows may become increasingly automated. Employees may move from generating work themselves to reviewing, validating or supervising work produced by AI systems.

Managers may become responsible for coordinating teams in which AI performs a substantial proportion of routine activity. Executives may face new decisions about investments, organisational structures, accountability and workforce planning.

As these responsibilities evolve, the competencies required to fulfil them will also change.

The UK government's AI Skills Adoption Pathway Model explicitly recognises this, setting out nine stages of organisational AI adoption and how skill needs change at each. The managerial research in the Strategic Management Journal provides complementary evidence that broader AI adoption is associated with changing demand for professional capabilities.

This suggests that AI capability benchmarking cannot be a one-time exercise. An organisation needs to understand its current competency distribution, define the capabilities required for its intended future state, and periodically reassess those requirements as its operating model evolves.

A different approach to building an AI-ready workforce

The collective evidence points toward a different model of organisational AI development.

Instead of beginning with a catalogue of AI courses and attempting to deliver those courses to as many employees as possible, organisations can begin by defining the competencies associated with the responsibilities their employees perform.

This involves understanding where AI is relevant to different business functions, where individuals exercise judgement or influence over AI-supported decisions, and where additional governance or oversight capabilities are required. From this, an organisation can establish its desired competency profile.

Assessments can then determine the extent to which the workforce currently meets those requirements, identifying gaps at individual, functional and organisational levels.

Learning interventions can be designed around those gaps, rather than around a universal expectation that every employee should acquire the same AI skills.

The effectiveness of those interventions should ultimately be evaluated through demonstrated competence and work outcomes, not simply course completion or self-reported confidence.

This approach preserves the importance of foundational AI literacy while recognising that higher-level proficiency, specialist applications and governance responsibilities need to be distributed according to organisational requirements.

It also provides a more meaningful definition of AI readiness. An organisation is not necessarily AI-ready because its employees are highly proficient in generic AI tools. It is AI-ready to the extent that its people possess the right combination of AI competencies to perform their responsibilities, exercise appropriate judgement and deliver the organisation's intended outcomes.

How the Kubicle AI Academy is built around this

This is the argument the Kubicle AI Academy was designed around. We do not train everyone toward a single standard of AI proficiency. We define four AI personas, each a different combination of capability rather than a rung on a ladder, and we measure people against the one their role actually calls for.

Four personas, four different shapes

Each persona has its own competency signature across four areas: AI theory, strategy and governance, architecture and engineering, and tools and platforms. The signatures are deliberately uneven, because the responsibilities behind them are.

A Navigator already needs the deepest tools and platforms capability, level with an Architect. An Architect needs relatively little strategy and governance. A Strategist needs the most strategy and governance, and comparatively little engineering. None of these is a more advanced version of another. In SFIA's terms, they are different competency requirements arising from different responsibilities.

Actual against desired, one person at a time

The Kubicle AI Literacy Assessment works in two steps, following the current-versus-target logic Benlian and Pinski recommend. The first step asks about a person's responsibilities, influence and accountability, and identifies the persona their role calls for: the desired capability. The second measures where their AI capability sits today across four pillars: the actual capability. The result shows both, and the gap between them, pillar by pillar.

Because the target comes from the role, the gap means something. A Strategist who scores highly on engineering has not closed a gap their role does not have, while a Navigator who is strong on theory but weaker on tools and platforms can see exactly where to focus. Assessing above the persona your role calls for is common and is no bad thing, but time is usually better spent going deeper in the capabilities your role depends on than reaching for a different persona.

Actual against desired, across the organisation

The same logic scales up. The AI Skills Navigator assesses a whole population and places each person in a persona, which gives an organisation its actual distribution of AI capability.

When a workforce is first assessed, people tend to land in broadly similar proportions: most as Apprentices (typically 40 to 60%), a smaller group as Navigators (25 to 35%), fewer as Architects (5 to 15%) and a handful as Strategists (2 to 8%). That is a picture of where people are, not of where they should be. The desired distribution is a separate question, answered by mapping each role to the persona it calls for, and it will differ from one organisation to the next for exactly the reasons set out above.

The difference between those two distributions is what the AI Academy is built to close, with learning pathways specific to each persona rather than a single catalogue delivered to everyone. And because the right mix shifts as AI adoption matures, reassessing at intervals shows whether the gap is genuinely closing.

A role-based approach also fits the direction of European regulation. Article 4 now asks providers and deployers to support AI literacy in a way that reflects people's knowledge, experience and context of use, rather than to certify a universal standard. Training each persona for the responsibilities it carries gives an organisation a clear, proportionate rationale for the measures it takes.

Conclusion: AI readiness is a question of alignment, not uniformity

The growing body of evidence from government, regulators, professional frameworks and academic research supports a consistent proposition: AI competencies are not universal, and their relevance depends on the responsibilities, context and influence of the individuals who possess them.

This establishes a strong conceptual basis for understanding AI capability as an organisational distribution rather than a single workforce-wide proficiency score.

The research does not yet establish a universally validated method for calculating the precise optimal competency distribution of every organisation. Nor does it imply that every company must have a completely unique profile.

What it does establish is that a uniform approach to AI competency development overlooks important differences in how people work, how decisions are made and how organisations create value.

The logical conclusion is that the desired distribution of AI competencies should be determined by an organisation's workforce composition, its responsibilities, its risk environment and its strategic ambitions. Its actual distribution reflects the capabilities employees currently possess. The difference between these distributions provides a more meaningful basis for diagnosing and developing organisational AI readiness.

The central question for organisations, therefore, should no longer be simply, "How AI-skilled is our workforce?" It should be, "Does our workforce have the AI skills it needs, in the places where they matter most?"

Next step: see how a managed program applies this across a whole workforce with AI training for organisations, from persona mapping to leadership reporting, or explore the full AI Academy and its four persona pathways.

References and further reading

1. Skills England (2025). AI Skills Tools Package, part of AI Skills for the UK Workforce. UK Department for Education, updated 4 November 2025. Includes the AI Skills Framework (entry, mid and managerial levels), the AI Skills Adoption Pathway Model and the Employer AI Adoption Checklist.
2. The Alan Turing Institute. AI Skills for Business Competency Framework. Supported by the UK Department for Science, Innovation and Technology and Innovate UK BridgeAI. Version 3 consultation draft published December 2025. Defines AI Citizens, Workers, Professionals and Leaders.
3. European Union (2024). Regulation (EU) 2024/1689 (the AI Act), Article 4: AI literacy, as amended by Regulation (EU) 2026/1744, published 24 July 2026.
4. European Commission. AI Literacy: Questions and Answers. Interpretative guidance on the Article 4 AI literacy obligation.
5. Velasco, L., Pal, S. and Schneider, C. (2026). An Operational Framework for AI Literacy in the Workplace. Interface, 25 August 2026. Proposes Baseline, Operational and Oversight tiers set by role and the consequences of AI use.
6. SFIA Foundation. SFIA: A Framework for AI Skills. Places AI skills within SFIA's seven levels of responsibility.
7. Benlian, A. and Pinski, M. (2025). The AI Literacy Development Canvas: Assessing and Building AI Literacy in Organizations. Business Horizons, published online October 2025. Introduces an AI Literacy Assessment Matrix and a development canvas for current versus target capability.
8. Hallman, N. J., Kowaleski, Z. T., Puvvada, A. and Schmidt, J. J. (2026). Sophistication in GenAI Use: Field Evidence from a Large Firm. Working paper, August 2026. Analysis of 713,564 prompts from nearly 4,000 back-office employees across 15 functional areas.
9. Alekseeva, L., Azar, J., Giné, M. and Samila, S. (2026). Artificial Intelligence Adoption and the Demand for Managerial Expertise. Strategic Management Journal. Examines the association between organisational AI adoption and demand for managerial skills.