AI has put middle management under the spotlight, but the real story may be Britain’s long-running accidental manager problem.

The logic appears simple. If AI can summarise meetings, generate reports, monitor workflows and move information between teams, why keep the management layers that used to do much of that work?

Where companies are cutting middle management roles while adopting AI, remaining managers are asked to take on larger spans of control – the so-called "player-coach" combining delivery, supervision and development in one stretched position.

Gartner predicted that through 2026, 20% of organisations would use AI to flatten structures, eliminating more than half of current middle-management positions — a shift that risks overwhelming remaining managers, increasing job insecurity and weakening mentoring pathways.

Britain already has a management capability problem hiding in plain sight.

CMI's Better Managed Britain report found that untrained managers often become responsible for their team's wellbeing, careers, productivity and compliance, despite having little preparation for that complexity.



That matters because AI doesn't arrive in a neutral organisation. It arrives inside existing cultures, reporting lines, habits and weaknesses. If a company already has managers who were promoted because they were technically strong, loyal or good under pressure, AI won't automatically make them better leaders. It may simply make the gap between task management and people leadership harder to ignore.

An accidental manager may be able to chase updates, attend meetings, pass on instructions and keep a team moving. But those are exactly the parts of management most exposed to AI tools. Meeting summaries, workflow tracking and progress reports are increasingly easy to automate.

The harder work sits elsewhere: diagnosing team dynamics, coaching people through ambiguity, managing conflict, protecting focus and helping good ideas survive organisational noise.

The danger is that organisations flatten the structure before they understand the function. They may remove a manager who mostly moves information around. That may be efficient. But they may also remove the person who notices when silence in a meeting means fear, not agreement.

When I searched “can AI replace managers?” Google gave me pages of results.

But that may be the wrong question.

Perhaps a better one is: which parts of management are real leadership, and which parts are admin in a leadership costume?

AI is already good at the work many people associate with middle management: summarising meetings, drafting updates, producing documents, analysing workflows and handling routine admin. GOV.UK research found that 85% of businesses using AI were using it for natural language processing and text generation.

That matters because much of modern management has become information-moving. Managers collect updates, attend meetings, chase progress, prepare reports and keep senior leaders informed.

Some of that work is useful. Some of it is just moving paper around the building.

AI is like a dishwasher in a restaurant kitchen. It can clean the plates faster, but it can’t decide what should be on the menu, notice that the team is overwhelmed, or challenge the head chef when the whole service is about to fall apart.

That’s the distinction leaders need to make.

AI can speed up the admin of management. But it can’t replace the judgement of leadership.

There is also a deeper risk. AI may not only automate parts of management. It may also make leaders more certain when they should be more curious. If a tool gives fast, confident answers that reinforce what a leader already thinks, it can make poor judgement feel efficient.

Leadership should not be reduced to getting a quicker answer.

Good leadership includes knowing when the answer needs to be questioned.

The third is human-signal management: noticing when silence means fear, when agreement is performative, when a high performer is close to burnout, or when a team has stopped challenging poor decisions.

The real issue isn't AI replacing middle managers. It's AI forcing leaders to separate management admin from leadership judgement. A meeting summary can be automated. The courage to say, "The team is nodding, but I know they don't agree" can’t.

The next generation of managers won't be valued for sitting between senior leaders and teams, passing information in both directions. AI tools and agents are already moving into that space. Microsoft's 2026 Work Trend Index argues that as AI agents take on more execution, leaders will need to redesign work around human and agent collaboration. The constraint is no longer only what people can do, but how work is structured around them.

That makes self-leadership a core AI leadership skill.

When managers lead larger teams, make faster decisions, and supervise AI-assisted workflows, their internal operating system comes into focus. Can they regulate their reactions under pressure? Stay clear-headed in ambiguity? Challenge a senior decision without becoming defensive? Notice when a team is complying outwardly but withdrawing emotionally?

Metacognition becomes a performance skill. Am I seeing the full picture, or just the version that confirms my frustration? Am I using AI to test my judgement, or to validate it?

These aren't soft skills. In an AI-enabled workplace, they become hard performance skills.

CIPD's Good Work Index found that stronger line manager ratings correlate with better performance, lower attrition, and greater openness around mental health. The more coordination AI handles, the more human leadership depends on judgement, emotional intelligence and trust. AI adoption should trigger a leadership development strategy, not just a technology one.

Used well, AI can support leadership development in practical, targeted ways. The key is to use it as a mirror, rehearsal space and decision-support tool, not as a replacement for judgement.

AI can help leaders:

• Personalise learning paths based on a manager’s role, strengths and development gaps, so learning feels relevant rather than generic.
• Simulate difficult conversations, including conflict, feedback, performance and change conversations, so leaders can practise before the stakes are real.
• Analyse meeting transcripts to highlight communication habits, missed opportunities, patterns of participation and examples of what a leader is doing well.
• Turn coaching notes into clear next actions, helping managers move from reflection to practical behaviour change.
• Prepare for key meetings by summarising background material, drafting agendas and surfacing decisions that need to be made.
• Spot patterns across feedback, engagement data or workflow information, giving leaders a clearer view of recurring risks or friction points.
• Test assumptions by asking what else might be true before a leader acts on their first interpretation of a problem.
  

Organisations should also be careful about which AI tools they use for which problems. AI can surface trends, risks and options from large amounts of information, but final judgement and accountability still remain with the leader.

A chatbot may help draft a project update. It may not be the right partner for diagnosing a trust breakdown, conflict pattern or team performance issue. The more human the problem, the more leaders need diagnostic tools, coaching and challenge that help them examine their own assumptions.

Before organisations use AI to flatten teams, widen spans of control or cut management roles, they need a clearer diagnosis of what is actually happening inside those teams.

The risk is treating every management problem as a productivity problem.

• Slow decisions become a process issue.
• Team silence becomes an engagement issue.
• Missed deadlines become a performance issue.

But beneath those symptoms may be something more specific: unclear priorities, low psychological safety, conflict avoidance, poor emotional regulation, weak capability or unrealistic workload. Those are different problems requiring different interventions.

A diagnosis-first approach asks leaders to pause before reaching for AI and answer five questions:

1. What problem are we actually trying to solve?
2. Are delays caused by unclear processes, or because people feel unsafe, confused or overloaded?
3. Are managers moving information around, or are they creating clarity, trust and accountability?
4. What evidence do we have beyond the loudest opinion in the room?
5. Which parts of this problem can AI support, and which parts still require human judgement?

This matters because AI can make leaders faster without making them wiser. Research published in Science found that sycophantic AI can reduce people’s willingness to repair interpersonal conflict while increasing their conviction that they are right. In a workplace, that could turn AI from a productivity tool into a conflict amplifier if managers use it to justify their frustration rather than examine their contribution to the problem.

A more useful starting point is to match the symptom to the likely cause.

If decisions are slow, ask whether the real issue is process, ownership or fear of making the wrong call.

If people are silent in meetings, ask whether they agree, have disengaged, or do not feel safe enough to challenge.

If managers are overloaded, ask whether AI can remove admin, or whether the organisation has simply designed an impossible role.

If teams are underperforming, ask whether the gap is capability, confidence, clarity or trust.

Only then should leaders decide where AI belongs.

AI may remove some administrative load from managers, but without stronger leadership capability, it may expose how little support they had in the first place. The answer is to use AI as a mirror, rehearsal space and decision-support tool, while strengthening the judgement, empathy and self-awareness managers still need.

Question for you: Is AI exposing weak middle management, or are leaders using it to avoid fixing weak leadership?

• Guardian: https://www.theguardian.com/technology/2026/may/15/ai-manager-purge-tech?utm_source=chatgpt.com
• ONS: https://www.ons.gov.uk/businessindustryandtrade/business/businessservices/bulletins/businessinsightsandimpactontheukeconomy/8january2026
• CMI: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.managers.org.uk/wp-content/uploads/2023/10/CMI_BMB_GoodManagment_Report.pdf
• Gartner: https://www.gartner.com/en/newsroom/press-releases/2024-10-22-gartner-unveils-top-predictions-for-it-organizations-and-users-in-2025-and-beyond