Contact
AI and Strategy

The People Blocking Your AI Programme Are Probably Your Middle Managers. Here Is Why That Is Your Fault.

When an AI programme slows down, there is a common diagnosis.

Middle management resistance. Hmmm….

The narrative writes itself. Leadership had a vision. The technology was sound. The data was ready. But somewhere in the middle of the organisation, people dragged their feet. They asked too many questions. They raised objections. They found reasons not to change.

Resistance. That is the word used.

Here is a different interpretation.

They were right to be cautious. And the problem was not them.

Key takeaways

  • Middle managers are the people who actually understand how work gets done. Their resistance is usually signal, not obstruction.
  • Most AI programmes ask middle managers to take on risk they did not agree to, for outcomes they do not control, with accountability they were not consulted about.
  • The failure is almost always in programme design and leadership behaviour, not in the people being asked to change.
  • Treating resistance as a communications problem produces better decks, not better adoption.
  • What works is involving the people closest to the work before the decisions are made, not after.

What middle managers actually know

Middle managers sit at a difficult intersection.

They are close enough to the work to understand what is actually happening: what the data really means, where the workflow breaks, what customers or patients or colleagues actually need, and why the last three initiatives did not land as planned.

They are also far enough from the top to have little influence over programme design, no control over timelines and limited ability to raise concerns without looking like they are obstructing progress.

When an AI programme arrives on their desk, they do a rapid and usually accurate assessment.

What is this actually going to change about how my team works?

Who is accountable if it goes wrong?

What happens to my people if this tool makes their role redundant?

Am I being asked to endorse something I do not understand well enough to defend?

Will my performance be measured differently once this is live?

Does this solve a problem my team actually has, or a problem someone at the top decided we have?

These are not obstructionist questions. They are responsible ones. The people asking them are doing their job.

The problem is that most AI programmes do not have good answers to any of them.

Why the answers are usually missing

AI programmes typically follow a familiar arc.

A senior team identifies an opportunity. A vendor is selected. A business case is built. A steering group is formed. Decks are produced. A pilot is run. Then the programme moves to rollout and suddenly needs the people it never consulted.

At that point, leadership notices something unexpected.

The people who must change their daily behaviour are not enthusiastic.

And because the programme did not anticipate this, it reaches for the standard tool: communications. A launch event. A roadshow. An FAQ document. A change management workstream staffed by people whose job is to get others to accept a decision they were not part of.

None of this works particularly well. People can feel the difference between being involved and being managed.

The result is compliance at best. At worst, it is a quiet, persistent refusal to fully commit: meetings attended, boxes ticked, tools logged into once, never used again.

Three things that actually drive adoption

  1. Involvement before the decision, not after

The people closest to the work should be involved in defining the problem, assessing the options and shaping how the solution is introduced, not presented with a rollout plan and asked to execute it.

This is slower at the front end. It is much faster overall, because it removes the adoption problem before it starts.

Middle managers who helped design the programme do not resist it. They defend it.

  1. Accountability that matches authority

One of the most reliable ways to create resistance is to hold people accountable for outcomes they do not control.

If a manager is expected to improve team performance using an AI tool but has no input into how that tool is configured, what data it uses, how errors are handled or how exceptions are managed, they are being set up to fail.

People who know they are being set up to fail will protect themselves. That protection looks like resistance.

Give people authority over the things they are accountable for. That is not a change management principle. It is basic organisational fairness.

  1. Leadership that removes obstacles rather than announcing vision

The most valuable thing a senior leader can do during an AI programme is not give a speech about transformation.

It is remove the obstacle that the team flagged three weeks ago and that nobody has fixed.

Middle managers are watching. They are trying to understand whether the organisation is genuinely committed or whether this is the latest in a long line of initiatives that looked important until the next priority arrived.

The signal they trust most is not the launch event. It is what leadership does when the programme gets difficult. Do they protect it? Do they resource it properly? Do they hold senior people accountable for blocking it?

Behaviour is the message. Everything else is noise.

The question worth asking

Before the next AI programme is diagnosed with a resistance problem, it is worth asking a different question.

Did we design this in a way that a thoughtful, experienced manager would be right to support?

Did we explain clearly what would change and what would not?

Did we involve the people closest to the work before the decisions were made?

Did we give people authority over the things they are accountable for?

Did our senior leaders behave as if this mattered when it got difficult?

If the answer to most of those is no, the programme does not have a resistance problem.

It has a design problem.

Fix the design. The resistance will largely take care of itself.

Who this matters for

This matters for executive teams, transformation directors, CIOs, programme sponsors and HR leaders managing large-scale AI or technology change. It is particularly relevant in complex organisations: healthcare systems, professional services firms, large public sector bodies, where professional identity, established practice and distributed accountability make top-down adoption strategies especially ineffective.

What leaders should do next

Before the next AI programme enters rollout, audit the design against three questions. Were the people closest to the work involved in shaping the solution? Does accountability match authority at every level? And have senior leaders demonstrated through behaviour, not communications, that this programme matters? If the answers are weak, address them before the rollout, not after adoption stalls.

author avatar
Richard A D Jones
Richard A D Jones is a healthcare AI strategy advisor, business strategy consultant, author and board-level AI adviser. He helps boards, CEOs, health-tech companies, investors and leadership teams understand where AI can create real value, where it creates risk, and how to turn strategy into execution. An author of books including Integrating AI into Strategy and Strategy Genius, Richard draws on experience across AI, healthcare, telecoms, startups, transformation and business model innovation. He has founded or supported 30+ startups, trained more than 2,000 executives, lectured at master’s level and spoken at major events on AI, strategy, innovation and digital health.

Richard A D Jones

Richard A D Jones is a healthcare AI strategy advisor, business strategy consultant, author and board-level AI adviser. He helps boards, CEOs, health-tech companies, investors and leadership teams understand where AI can create real value, where it creates risk, and how to turn strategy into execution.

An author of books including Integrating AI into Strategy and Strategy Genius, Richard draws on experience across AI, healthcare, telecoms, startups, transformation and business model innovation. He has founded or supported 30+ startups, trained more than 2,000 executives, lectured at master’s level and spoken at major events on AI, strategy, innovation and digital health.

Leave a Reply

Your email address will not be published. Required fields are marked *

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Youtube
Consent to display content from - Youtube
Vimeo
Consent to display content from - Vimeo
Google Maps
Consent to display content from - Google
Spotify
Consent to display content from - Spotify
Sound Cloud
Consent to display content from - Sound