KPMG: AI made expertise abundant – here's what became scarce instead

For decades, organisations built competitive advantage by hoarding expertise. They hired specialists, retained consultants and constructed entire departments around people who knew what others did not. Access to expert knowledge determined who won contracts, shipped products first or advised the most valuable clients. That economy is ending.

At the Nrth Festival in Toronto, a KPMG transformation partner argued that generative AI has flipped the scarcity curve. Expert knowledge – the kind that once required years of apprenticeship, professional accreditation or expensive advisory retainers – is now abundant. Large language models can draft legal clauses, debug code, summarise clinical trials and propose market-entry strategies in seconds. The bottleneck has moved. What organisations lack today is not expertise but the judgment to apply it, the curiosity to ask better questions and the organisational courage to act on answers that contradict legacy assumptions.

This shift carries immediate stakes for L&D and talent leaders. If your training programmes still optimise for knowledge transfer – teaching people what AI can already retrieve – you are solving yesterday's problem. The new curriculum must centre on the capabilities that remain scarce when expertise becomes a commodity: critical thinking under ambiguity, the ability to frame problems that machines cannot yet recognise and the interpersonal skill to build trust when algorithms generate the first draft.

What abundance changes

When expertise was scarce, organisations competed by locking it inside full-time headcount or exclusive advisory relationships. A specialist's value lay in what they knew that you did not. Generative AI collapses that information asymmetry. A junior analyst with access to a well-prompted model can now produce work that once required a senior consultant's billable hours. The model does not replace the consultant's judgment, but it does eliminate the need to pay for routine knowledge retrieval dressed up as bespoke insight.

KPMG's partner pointed to a second-order effect: when everyone can access expert-level outputs, differentiation moves to the questions you ask and the context you provide. A legal team that uses AI to draft contracts faster gains little advantage if every competitor does the same. The team that uses AI to surface edge cases in contract language – because someone asked a sharper question – creates distance. The scarcity has shifted from answers to the quality of inquiry.

This reframes what learning and development must deliver. Traditional training often prioritised declarative knowledge: teach someone the tax code, the project-management framework or the compliance checklist. That knowledge is now table stakes, available on demand. The new training challenge is procedural and metacognitive: teach people how to interrogate an AI-generated answer, how to recognise when a plausible output is subtly wrong and how to decide which problems are worth solving in the first place.

The new scarcities

If expertise is abundant, three capabilities become the new competitive constraints. The first is contextual judgment. AI can generate a marketing plan, but it cannot know that your executive team will reject any proposal that cannibalises the legacy product line. It can draft a restructuring memo, but it cannot sense that the workforce will interpret "efficiency" as a euphemism for redundancies. Humans who understand organisational context – the unwritten rules, the political terrain, the cultural antibodies – become the editors who turn plausible AI outputs into implementable strategy.

The second scarcity is question design. Generative models are only as useful as the prompts they receive. A vague question yields a generic answer. A precise question that incorporates constraints, success criteria and edge cases yields a tailored solution. The ability to frame problems clearly, to specify what good looks like and to iterate on a prompt until the output is fit for purpose is a skill that most organisations have not yet systematised. L&D leaders who treat prompt engineering as a technical curiosity rather than a core literacy are leaving performance on the table.

The third scarcity is the willingness to act on insight that contradicts institutional inertia. AI can surface patterns in customer data that suggest your flagship product is losing relevance. It can model scenarios that show a faster path to market if you abandon a legacy process. But organisations do not fail because they lack insight; they fail because insight threatens someone's budget, someone's status or someone's mental model of how the business works. The scarcity is not analytical horsepower. It is the organisational courage to act on what the analysis reveals, even when it is uncomfortable.

What this means for L&D

Learning and development teams that continue to design programmes around knowledge transfer will find themselves training people for a world that no longer exists. The new curriculum must prioritise three shifts. First, move from teaching answers to teaching inquiry. Train people to ask second- and third-order questions, to probe assumptions and to recognise when an AI-generated answer is plausible but incomplete. Second, embed judgment training into every workflow where AI is deployed. Do not assume that people who can use a tool will automatically know when to distrust it. Third, create forums where teams can practice acting on uncomfortable insight without career risk. If your organisation punishes people for proposing ideas that challenge the status quo, no amount of AI access will matter.

KPMG's argument is not that expertise has no value. It is that the value has migrated. The consultant who knows the tax code is now competing with a model that knows it faster. The consultant who knows which tax strategy will survive an audit, align with the client's risk appetite and avoid a board-level fight is still irreplaceable. L&D's job is to close that gap – not by teaching more facts, but by building the judgment, curiosity and courage that remain scarce when expertise is everywhere.

Sources:

• Human Resources Director, 29 September 2026