The CEOs’ numbers are more honest than their narrative

The essence

• IBM Institute for Business Value surveyed 2,000 CEOs worldwide on how AI is transforming their organisations. The messaging around the report sounds confident; the figures themselves tell a more uncomfortable story.

• The problem lies not only in employees’ capabilities, but also in how work around the technology is organised.

Ask two thousand CEOs whether their employees have the skills required to work alongside AI, and 86 per cent say yes. Ask them how many employees are actually using AI in their day-to-day work, and the answer drops to 25 per cent. Those two figures appear side by side in the same report: the 2026 CEO Study by the IBM Institute for Business Value, published in early May and based on a global survey of 2,000 CEOs.

The narrative surrounding the report sounds self-assured. AI-first organisations are said to have achieved 17 per cent higher revenue growth over the past three years. Seventy-six per cent of the companies surveyed now have a Chief AI Officer, compared with 26 per cent a year ago. By 2030, the CEOs surveyed expect AI to autonomously make almost half of all operational decisions.

Humans won’t disappear from the loop. Instead, their role will shift from making each decision to designing the decision logic, setting guardrails, and stepping in only when exceptions carry material, ethical, or strategic consequences.  

-  2026 CEO Study by the IBM Institute for Business Value

But when the same report is read through the eyes of an employee rather than a CEO, a different picture emerges.

The gap between what employees are supposedly capable of according to their CEOs and what they actually do is captured by the report itself in one sentence: this is “more an organisational design problem than a skills problem”. The cause becomes immediately clear. The bottleneck lies in the way work is structured around the technology, not with employees who are unwilling or unable to use it.

Between 2026 and 2028, CEOs expect that 29 per cent of employees will require reskilling for a new role, while 53 per cent will need reskilling for their current role. That is a substantial undertaking for a challenge which, according to the same report, is not merely a training issue.

The report itself points towards the solution: redesign the organisation before the reskilling begins. According to IBM, organisations that systematically redesign five core areas — technology, finance, HR, operations, and cross-functional collaboration — are four times more likely to achieve the objectives set out in their business case.

The first step is to define clearly which decisions, judgements, and handovers remain human, and which do not. More technology alone is not enough. The sequence matters: optimise, standardise, automate.

Today, AI-powered process mining should first and foremost help organisations identify friction points. Through AI, organisations can now map what sampling methods could never fully reveal: how work actually flows in practice, how often processes follow the ideal path, or happy flow, and where exceptions occur.

Once that picture is clear, organisations can consciously decide which processes to standardise, which elements to automate using AI flows or agents, and which forms of judgement should remain human.

Without that preparation, organisations risk scaling up the very frictions, exceptions, and workarounds they have been manually compensating for over many years. Those that genuinely achieve efficiency gains do so by deploying AI deliberately and at the right point in the chain.

That deliberate deployment requires the right people around the table. In the space of a single year, the share of organisations with a Chief AI Officer has tripled. One hundred per cent of CEOs expect the influence of the CAIO to continue growing in the years ahead. For the Chief Human Resources Officer, that figure stands at 59 per cent.

Appointing a CAIO is a positive step because it puts AI high on the strategic agenda. But embedding AI into processes, skills, and competencies is work led by HR and operations. A CAIO sprinting alongside the CEO may move quickly at executive level, but that alone does not guarantee implementation on the workfloor.

The organisations that will succeed are those in which the CEO, CAIO, CHRO, and COO steer the transformation together.

IBM summarises this in a single line in its own recommendations on page 36: redesign the workflow first, and only then redesign the jobs. Today, 87 per cent of surveyed CEOs say they are actively embedding AI into workflows. Yet only a quarter of employees are actually using AI in practice.

Organisations aiming for 17 per cent higher revenue growth should therefore start by redesigning the organisation — not merely rewriting employees’ job descriptions.