Middle managers are failing AI adoption because organisations aren't equipping them to succeed
78 per cent of managers now use AI regularly, up from 64 per cent two years ago. But frontline worker adoption has stalled at 51 per cent. The gap between those numbers reveals the real problem: managers are adopting AI themselves while lacking the system, guidance and authority to drive adoption in their teams.
This is where AI strategy dies. Leadership sets direction from the top. Frontline workers feel the impact at the bottom. But the middle managers who translate one into the other have been left out of the conversation entirely.
The middle layer does more than pass information
Gartner predicts that by the end of 2026, one in five organisations will use AI to flatten their structures, eliminating more than half of current middle management positions. The logic is straightforward: AI surfaces information directly and automates decisions that once required human aggregation. The middle becomes overhead.
This misreads what middle managers actually do. The ones worth keeping interpret context, make judgment calls about priorities, absorb organisational friction and translate abstract direction into concrete action. These are precisely the capabilities AI cannot replace-and precisely what organisations will need more of as AI reshapes how work gets done.
The question is not whether middle management is redundant. It is whether organisations will equip the people in those roles for what the role is becoming.
Managers adopt AI. Their teams do not.
A Cornerstone OnDemand survey from late 2025 found that 80 per cent of employees now use AI at work. Yet 57 per cent are reluctant to tell their manager they do so. Only 44 per cent have received any AI training at all.
Some employees fear judgment from colleagues. Others lack guidance on what good AI use looks like. In both cases, nobody has created the conditions for them to use AI openly and well. Their managers have not been given the mandate, training or framework to do it.
AI mandates stall when the middle is bypassed
Research published by TechPolicy.Press in early 2026, based on interviews with 50 middle managers, found a consistent pattern across industries. The mandate lands. The tools are provisioned. Completion metrics are tracked. Then nothing changes.
Managers are under pressure to demonstrate AI success to leadership while simultaneously lacking the resources to deliver it. The result is what researchers call an illusion of a successful AI rollout-managers perform adoption for their superiors without embedding it in how their teams actually work.
The organisation asked managers to deliver something it never equipped them to deliver. That is a failure of design.
The manager's role is shifting from information broker to capability builder
The traditional middle management role centred on gathering information, aggregating it, reporting upward and distributing decisions downward. AI compresses most of that. What grows is the work of building capability in the people around you.
That means helping a team member figure out where AI fits in their specific workflow. It means knowing when AI-assisted output is good enough and when it needs human review. It means creating conditions where people learn by doing real work with AI, not by completing a course about it.
Very few AI training programmes address this. Very few management systems capture and share what people learn on the job.
AI strategy should start with the middle, not above it
Organisations that want AI to move beyond basic adoption and abandoned pilots need a different approach. The people who will make AI work are the managers in between-the ones who shape how teams work every day.
Investing in this layer means giving managers the authority and support to work with their teams on how AI enters the work. It means building capability through practice rather than instruction. It means making the results visible and accountable.
Where AI actually lands inside most organisations will depend on the layer in between. Right now, most organisations are not even talking to them. Consider exploring AI for Management or AI Learning Paths for Managers to understand how to equip your middle management layer for this shift.
Your membership also unlocks:
