I’ve come to believe that the biggest barrier to AI adoption is rarely whether people can learn the technology.
It is whether they actually want to change how they work.
That sounds like a subtle distinction, but in practice it is enormous. A business can invest in the best AI tools, give employees access to them and run technically excellent training programmes and still see very little change in day-to-day behaviour.
The technology can be ready long before the organisation is.
The Technology Is Not Usually the Hardest Part
When a new AI model is released, the conversation tends to focus on capability.
How accurate is it? How fast is it? Which model should we use? Can it connect to our existing systems?
Those are important questions. But once you move from experimentation into an organisation, a different set of questions becomes much more important.
Will employees actually use it?
Will managers encourage them to use it?
Will people trust the output enough to change their workflow?
Will they feel comfortable experimenting and making mistakes?
And perhaps most importantly: what happens when using AI requires people to change a habit they have built over years?
That is where behavioural psychology becomes relevant.
AI adoption is not simply a technology implementation. It is a change in behaviour.
Microsoft and the Power of Habit
One of the clearest examples comes from Microsoft’s own research into Copilot adoption.
Microsoft studied 1,300 Microsoft 365 Copilot users and found that saving just 11 minutes a day was enough to reach what it described as an “11-by-11 tipping point” the point at which users began to perceive meaningful value from AI.
I find that interesting because it is not really a technology story.
It is a habit story.
An employee does not need to transform their entire working life on day one. They need to experience one small, repeatable improvement that makes the new behaviour worthwhile.
Once that behaviour becomes familiar, it can expand.
That is very different from telling 500 employees, “Here is your new AI tool. Please use it.”
The first approach creates experience.
The second creates a licence.
One of the clearest examples comes from Microsoft’s own research into Copilot adoption.
Microsoft studied 1,300 Microsoft 365 Copilot users and found that saving just 11 minutes a day was enough to reach what it described as an “11-by-11 tipping point” the point at which users began to perceive meaningful value from AI.
I find that interesting because it is not really a technology story.
It is a habit story.
An employee does not need to transform their entire working life on day one. They need to experience one small, repeatable improvement that makes the new behaviour worthwhile.
Once that behaviour becomes familiar, it can expand.
That is very different from telling 500 employees, “Here is your new AI tool. Please use it.”
The first approach creates experience.
The second creates a licence.
Accenture: Skills Are Only Part of the Equation
Accenture offers another useful example.
In fiscal 2025, the company invested around $1 billion in learning and professional development, delivered approximately 47 million training hours, and reported that more than 550,000 employees had completed training on generative AI fundamentals. It also had around 77,000 skilled AI and data professionals.
The scale is impressive.
But what stands out to me is that Accenture does not treat AI capability as simply a matter of giving people a course. Its programmes include role-based boot camps, practical learning and opportunities to apply AI in real work.
That distinction matters.
Knowing how a tool works is one thing.
Knowing when to use it, trusting it appropriately, knowing when to challenge it and making it part of your normal workflow is something else entirely.
Technical skill gives someone the ability to use AI.
Behaviour determines whether they actually do.
Take My Own Experience at Fliweel
This is something I have become increasingly aware of through my own work at Fliweel.
When I started Fliweel, much of the conversation around automation was understandably technical. Businesses wanted to know what could be automated, which platforms to use and how much time or money a solution could save.
Those questions still matter.
But the more organisations I work with, the more I notice that the difficult part often comes after the technology works.
You can build an AI assistant. You can connect the systems. You can make the workflow reliable.
Then someone still has to change the way they work.
One example is our work with the Electrical Contractors’ Association on AskECA, an AI assistant trained on more than 70 technical documents to support live member helpline calls. The technical architecture was important, but so was understanding how people would actually use the system in a real conversation.
Testing uncovered an important limitation: the system could become too eager to provide an answer when repeatedly questioned. As Daniel Bailey from ECA put it:
“AI really wants to please you. Repeated questioning eventually breaks guardrails.”
That observation shaped the way the system was used.
Instead of allowing the AI to generate freely, ECA kept the assistant retrieval-based so that it remained grounded in its approved technical documentation.
For me, that is a good example of the relationship between technology and behaviour.
The system had to be designed around how people actually interact with it — including the ways they might test it, challenge it or misunderstand its limitations.
Why Trust Changes Behaviour
This is where behavioural psychology becomes particularly important.
People do not adopt new technology simply because somebody proves that it works.
They adopt it when they believe that using it is safe, useful and worth the effort.
There is also a psychological cost to changing a familiar workflow.
If someone has spent ten years developing expertise in a particular process, asking them to let an AI system participate in that process can feel like more than a productivity change. It can feel like a challenge to their judgement or professional identity.
That is why trust matters.
Recent McKinsey research found that employees are generally more ready to use AI than their organisations are ready to change around it. The research also found that trust in an organisation’s support during AI-related change is associated with greater enterprise value across stages of AI maturity.
That tells me something important.
If people are worried about losing control, making mistakes or becoming less valuable, giving them another technical training session will not necessarily solve the problem.
They need clarity.
They need visible leadership.
They need permission to experiment.
And they need to understand where human judgement still matters.
The Real AI Adoption Challenge
I don’t think this means technical skills are unimportant.
They are essential.
But technical capability is becoming increasingly accessible. AI tools are easier to use, interfaces are improving and people can learn the basics remarkably quickly.
The harder advantage to replicate is an organisation that knows how to change.
That means creating an environment where people:
-
Experiment without being afraid of failure
-
Understand why AI is being introduced
-
See leaders using the technology themselves
-
Know what good AI-assisted work looks like
-
Understand when human judgement is required
-
Share successful use cases with colleagues
-
Have time to build new habits
That is behavioural change.
And behavioural change is much harder to achieve by simply buying software.
What I’ve Learned
If I could give one piece of advice to a business starting its AI journey, it would be this:
Don’t measure your AI strategy by how many people have access to the technology. Measure it by what people are actually doing differently.
A company can have 1,000 AI licences and still have an adoption problem.
Equally, a team of 20 people using AI consistently to remove friction from their daily work can create significant value.
The question I would ask is not:
“Do our people have the technical skills to use AI?”
I would ask:
“Have we created the conditions that make people want to use it?”
That changes the conversation completely.
The future of AI adoption will belong not simply to the organisations with the best technology, but to those that understand the people using it.
Because ultimately, AI does not transform a business on its own.
People do.
How Fliweel Can Help
If your organisation is investing in AI but adoption is proving harder than expected, the answer may not be another tool or another training session. It may be time to look more closely at the people, processes and behaviours surrounding the technology.
At Fliweel, I help organisations move from AI experimentation to practical implementation, identifying where AI can create genuine value, understanding the barriers to adoption and building solutions that people can actually use.
If you’re considering where AI could make the biggest difference in your organisation, get in touch with Fliweel to start the conversation.
ABOUT FLIWEEL.TECH
Fliweel.tech is a leading provider of AI and automation solutions, specialising in intelligent bot development and robotic process automation. Our mission is to help businesses streamline their operations, reduce errors, and focus on higher-value tasks through innovative technology. With a commitment to excellence and customer satisfaction, Fliweel.tech delivers customised solutions that drive tangible results for clients across various industries.
GET IN TOUCH
Learn how AI and automation can transform your business