Organisations Keep Mistaking Technology Access for Capability Creation
Countless large companies are making the same bet right now: buy enough high-powered AI licences, distribute them widely, and adoption will follow. The prevailing belief is that adoption will drive productivity, which will justify the spend, making the organisation more innovative along the way.
While the logic feels obvious enough that few stop to examine it, both assumptions are questionable. AI is expensive today, and despite expectations to the contrary, the long-term direction of travel may not be downward. Many AI providers are absorbing enormous infrastructure costs and will eventually need to recover them. More importantly, the foundational assumption, that access creates adoption and adoption creates transformation, is one we've already tested more than once.
The results were not especially encouraging.
We've run this experiment before
In the 1990s, Microsoft Office put a macro recorder on every desk with a compelling promise: anyone could automate repetitive work without becoming a software developer. In theory, millions of office workers would eliminate mundane tasks and become dramatically more productive. In practice, only a small number of people embraced it. They built the macros, automated the reports, and created the little tools that saved everyone else an hour every month, while most people never touched it at all.
This pattern repeated itself in the 2010s with low-code platforms. Products such as Power Platform arrived with a similar proposition of drag-and-drop development, empowering business users to become 'citizen developers.' Organisations invested heavily and provisioned licences widely in anticipation of broad adoption. A decade later, while some employees genuinely became builders, creating useful applications and automating meaningful chunks of work, the vast majority never did. The capability existed, the tools worked, and the licences were available, yet the scale of adoption assumed in the investment case failed to materialise.
Generative AI is undoubtedly a bigger leap forward than either of those technologies, lowering the barrier to building further than ever before. A person can now create software, automate workflows, analyse data, or generate content using natural language rather than code. What has not changed nearly as much, however, is the distribution of human behaviour. The people who enjoyed building before still enjoy building now. The people who actively seek opportunities to improve and automate their work remain a minority. Ultimately, AI makes builders more capable, but it does not necessarily create more builders.
AI amplifies existing behaviour
In most teams, a small subset of individuals will take a new tool and immediately start experimenting. They will automate recurring tasks, build internal tools, create workflows, and find ways to eliminate work they find repetitive or frustrating. Give them a powerful AI licence, and they will almost certainly create value.
However, that value is often highly local. Builders naturally focus on the problems closest to them; they know where their own time is wasted and which daily processes frustrate them. What they often lack is visibility across the wider system. They cannot easily see where the organisation is losing millions while saving themselves a few hours. This isn't because they lack intelligence, ambition, or creativity, but because their role rarely requires that macro perspective. Most people are hired to perform within a system, not to redesign it.
This is why widespread AI adoption often produces a collection of useful improvements, rather than meaningful organisational change. The initial investment is justified using the language of transformation, but the actual benefits arrive merely in the form of convenience.
What actually changes a business
- Business transformation requires much more than simply building capability. It requires several kinds of people:
- People who can identify where value is being lost: professionals who understand processes, incentives, customer journeys, operating models, and organisational friction well enough to recognise opportunities that genuinely matter.
- People who can build the solutions.
- People with the authority and influence to change the way the organisation functions.
AI dramatically strengthens the second group, allowing builders to execute faster, experiment more cheaply, and solve increasingly complex problems. What it does not automatically create, however, is the first or the third. A powerful AI tool can help someone implement a solution, but it cannot tell them which problem is most worth solving. It cannot align stakeholders, redesign incentives, or drive behavioural change across an organisation. Those critical capabilities remain stubbornly human.
This is where many AI strategies begin to unravel. Organisations invest heavily in capability and assume that problem identification and business change will naturally emerge alongside it, mistaking technology access for capability creation. They have made this same mistake before: the macro recorder did not turn everyone into a builder, low-code platforms did not create armies of citizen developers, and distributing AI licences will not create armies of transformation specialists.
AI will undoubtedly help people build more things, but whether those things actually matter to the business is a different question entirely. The primary challenge has never been a shortage of tools; it is knowing where technology can genuinely change the economics of the business and having people willing and able to act on that insight.
Neither of these problems will be solved simply by buying more licences.
