All insights

Beyond the Tech: Why AI adoption fails and what leaders are missing

By Emma Rose Hurst

Many organisations believe AI adoption is a technology challenge. In reality, it’s a people challenge. The difference between organisations that see real value from AI and those that struggle often comes down to clarity, confidence, and culture rather than capability.

Across sectors, we are seeing significant investment into AI tools, with new platforms being deployed, licences being purchased, and pilots being launched at pace. Yet increasingly, organisations are encountering the same issue: buying AI does not mean people will use it well.

When people are not properly equipped, adoption fails, but not necessarily in obvious ways. Instead, it shows up more quietly through hesitation, inconsistent, unproductive and unsafe use, and uncertainty about what good looks like. Over time, this hugely limits the value organisations hoped to achieve.

The real risk is not the technology

One of the clearest patterns we are seeing is that organisations are focusing far more on the technology than on the experience of using it. If we’re lucky, training is delivered, tools are introduced, and expectations are set, but training alone does not create readiness.

To use AI effectively, people need more than technical instruction. They need to understand why it is being introduced, what is expected of them, where the boundaries sit, and how far they are trusted to experiment. They also need managers who can help them navigate the uncertainty that comes with trying something new.

Without that broader support, adoption becomes fragmented and inconsistent, even when the technology itself is sound.

People may be using AI, but that does not mean they trust it

This became clear during our recent webinar with HR, transformation, and digital leaders. When we asked whether people in their organisations felt safe, confident, and in control when using AI, around 60 percent selected the same response:

People were using AI when told to, but they did not fully trust it and felt their concerns were not properly addressed.

That insight is significant. Usage alone does not equal adoption. People can comply with demands to use a tool a certain amount while still feeling uncertain about what they are doing. When that happens, behaviour tends to be cautious and inconsistent. Morale declines. Questions remain unspoken, risks go unnoticed, and opportunities are missed.

Confidence is not created by instruction alone. It is created by clarity, transparency, and visible leadership support.

In the absence of clarity, people create their own rules

When organisations fail to define what good AI use looks like, employees rarely stop and wait for guidance. Instead, they begin to improvise.

This is where shadow use happens. People test tools on the sly, use personal accounts and develop their own ways of working because they feel under pressure to use AI but don’t know how to do so in accordance with their organisations wishes.

We also see resistance. While a third of people admit to actively sabotaging their organisations AI efforts, most of this resistance is not in the form of open refusal, but hesitation and minimal engagement. From the outside, it may appear that adoption is progressing, yet internally, trust may be fragile, and behaviour inconsistent.

Over time, that gap between appearance and reality becomes a source of risk.

The ownership gap is slowing progress

Another issue frequently slowing adoption is a lack of responsibility.

Leadership assumes adoption will follow investment. IT focuses on systems and infrastructure. HR focuses on capability and behaviour. No one takes responsibility for building a pro-AI culture. Managers are left trying to interpret what AI means for their teams and individuals are left flailing.

We saw this reflected in the second poll during the webinar. When asked whether organisations knew which tasks should be done by people, AI, or a mix of both, around 60 percent of participants indicated that people were largely working it out for themselves, with no shared approach in place.

That is not sustainable adoption. It is a signal that technology has been introduced, but expectations have not been clearly defined.

AI adoption is a change challenge, not just a tooling rollout

This is one reason AI adoption cannot be treated like a standard system implementation. In the past, system implementations have been seen as defined programmes with clear endpoints: launch the system, train the workforce and move into steady state of use. However, AI does not behave that way. Tools are evolving rapidly, expectations shift, and new possibilities are constantly emerging. What leaders and organisations are managing now is not a one-and-done implementation, but an ongoing process of change.

This therefore requires a shift in mindset. Leaders and managers need to think less about delivering a single transformation and more about sustaining continuous evolution. Training cannot be a one-time event, communication cannot be limited to launch moments, and embedding must become an ongoing discipline rather than a one-off activity.

Where adoption usually gets stuck

In many organisations, the biggest challenge is not getting started; it’s making progress once the tool(s) have gone live.

AI adoption typically stalls when organisations move from awareness to practical use. Leaders announce initiatives and sometimes encourage experimentation, but fail to define where AI genuinely adds value, where human judgement must remain central, and how success should be measured.

Without that clarity, experimentation begins to feel risky rather than productive. People hesitate, second-guess their decisions, and worry about making mistakes. If the environment does not make learning, experimentation and failure safe, progress slows and confidence declines.

This is often the point where organisations realise that technology alone will not deliver the results they expected.

The organisations that benefit most will prepare their people properly

The organisations that gain most from AI will not necessarily be those with the most advanced tools. They will be the ones that invest equally in preparing their people.

They will build trust early, create clear guardrails, support managers to lead conversations within their teams, and make experimentation both safe and purposeful. They will recognise that adoption is not simply about encouraging use, but about enabling confident and consistent behaviour.

For many organisations, the most effective first step is smaller than expected. It is not a large-scale transformation programme, but a period of discovery, reflection and insight. Understanding where AI is already being used, where confusion exists, and where risks are emerging provides leaders with the clarity they need to move forward with confidence.

Technology may accelerate capability, but culture determines whether that capability translates into performance. The organisations that are pulling ahead today are not simply investing in AI. They are investing in the clarity, confidence, and behaviours that make AI usable in practice.

A practical next step for leaders

If there is one lesson emerging consistently, it is this: successful AI adoption begins with understanding, not implementation.

Before launching the next tool or initiative, leaders should take time to understand how AI is currently being used across the organisation, where uncertainty exists, and what support people genuinely need.

That insight provides the foundation for everything that follows. Without it, adoption becomes reactive. With it, adoption becomes intentional.

If you missed it, you can find the recording here.