In any large organisation that has been talking about artificial intelligence for the last three years, there is a meeting that takes place once a quarter. It has a different name each time. Two years ago it was the digital transformation review. This year it is the AI transformation update. The deck is new. The consultants are slightly more expensive. The vocabulary has acquired some technical inflections it did not have a decade ago—the current buzz is "headless AI", a phrase uttered without any sense of irony. A slide is shown on which the words "human-centred" appear in serif type alongside the words "data-driven" in sans-serif, separated by a suitable graphic. The steering committee nods. The workstream leads report green. The chief transformation officer assures the chief executive that the programme is "on track".
What none of those present have done, in the period since the previous meeting, is materially redesign any of the work the transformation programme is meant to be transforming. The slides have changed. The job descriptions have not. The all-hands script has changed. The org chart has not. The annual report has acquired a new sub-section. The performance management system has not. The transformation is operationally undetectable. It is, however, audible at considerable volume in every executive forum in the organisation.
This is the transformation gap, and it is the cleanest evidence available that the vocabulary of organisational change has, over the past decade, come unmoored from the practice of it. The Deloitte survey numbers provide the empirical confirmation—thirty-four per cent of organisations report deeply transforming their businesses through AI; eighty-four per cent admit no redesign of jobs around AI capabilities—but the operational diagnosis does not require them. The meetings tell us. The slides tell us. The annual reports tell us. The work, which has not changed, tells us most of all.
The disconnect is not deliberate. The executives reporting deep transformation are not lying in any narrow sense; they have not invented the strategy decks or the kick-off meetings or the transformation office on the third floor with its dedicated programme management function and its quarterly KPIs measuring percentage-of-employees-AI-trained. All of that is real. It has only stopped corresponding to anything operationally consequential. The performance has become detached from the work the performance was supposed to be performing. The underlying mechanism was described almost fifty years ago: organisations adopt formal structures whose primary function is to signal legitimacy to external audiences, regardless of whether those structures bear any operational relation to the work the organisation actually does.