Why successful AI transformation starts with people

Public sector organisations are under growing pressure to harness AI to improve services, increase efficiency and free employees from repetitive work. But deploying the technology is only part of the challenge. Turning AI pilots into lasting transformation means understanding where it can genuinely add value, bringing employees with you and being clear about what success looks like.

Why is it important to start with people and organisational needs when thinking about AI transformation?

AI transformation isn’t primarily a technology problem. The technology exists. It’s when you align it with organisational needs, and the needs of citizens and colleagues, that it starts to add real value.

If you start with the technology, you might end up with something really clever, but what does it actually deliver? If you start with the organisational need and people first, you can identify the critical outcomes the organisation needs and focus on those.

We see a lot of proofs of concept failing, and part of the reason is that organisations start with a technology-first mindset. You should start with your people and transformational objectives, then address the problems that will deliver the return on investment and impact you’re looking for.

Why is moving from AI pilots to widespread adoption so difficult?

A pilot or proof of concept is relatively simple. When you move to implementation at scale, however, it needs to stick within the organisation – and that isn’t just about technology.

People need to understand what they should use AI for, what they are and aren’t allowed to do, and the role they play. You also need governance and assurance, clear ownership, the right skills and strong leadership support.

We sometimes describe this as the “missing middle”. You have the ambition and technology at one end and the expected benefits at the other. What can be missing is everything required to make AI genuinely adoptable: people, skills, governance, ownership and communication.

Why do AI programmes sometimes struggle to gain traction with employees?

For people to adopt AI, it needs to feel relevant, safe to use and valuable to what they are doing. They also need to feel supported.

Otherwise, people can worry about making mistakes or feel that change is being done to them rather than with them. AI might be new, but we shouldn’t forget what we’ve learnt from previous technology transformations. If you don’t bring people with you, it’s difficult to gain traction.

We’ve seen this through our work with Shared Services Connected Ltd (SSCL), where we have been looking at how to make the organisation more AI-native.

We didn’t start with technology. We started by setting a vision for what AI meant for the organisation and educating the leadership team about what it can do, as well as the potential risks. We then engaged with the workforce, asking what worried them about AI and what concerns needed to be addressed.

By the time you start talking about the technology itself, people are less fearful and more excited about what it could do for them.

How can organisations build trust and encourage employees to embrace AI?

Engage people from the beginning. Employees need to understand what AI can do and its limitations, but also the role they continue to play.

Human judgement remains incredibly important. Critical thinking and human judgement are still essential, and if employees understand that they remain critical to the process, it can help them embrace AI.

Frontline employees can also have the best view of where AI could deliver value. If you’re looking for operational efficiencies or reimagining how work gets done, involve the people who understand how the organisation operates day to day.

The worst outcome is to put significant time, resources and money into something and then discover it won’t solve the problem you thought it would. Involving employees early makes that less likely.

How can organisations tell whether AI adoption is actually delivering value?

This is one of the big questions. Organisations might say lots of people are using Copilot or another AI solution and everybody is saving 20 minutes a day. But what does that actually mean from a CFO’s perspective?

You need to be targeted about the outcomes you intend to achieve and clear about how you will measure them. If you say AI is going to create value but can’t measure that value, how do you know whether you’ve achieved the outcome?

Not every measure needs to be financial. It could be employee satisfaction, customer satisfaction or another measure relevant to the organisation. But you need to define and actively track it.

Simply knowing somebody has logged into an AI system doesn’t tell you much. Are you delivering outputs faster? Responding to citizens or customers more quickly? Are satisfaction levels improving? Are efficiencies allowing employees to focus on more critical activities? Those measures tell you whether AI is genuinely creating value.

What does a human-centred approach to AI look like in practice?

Being human-centred helps mitigate many of the risks we’ve discussed. People are often an organisation’s most important asset, and they should be treated as such when it comes to AI.

That means giving people a clear vision and leadership, providing the skills they need, putting governance around what they are doing and supporting the behavioural change required.

You can almost flip the idea of “human in the loop” around. Put the human at the centre and have AI in the loop. AI can do a lot of the heavy lifting and bring scale, but critical thinking and judgement still sit with people.

Without that judgement, organisations are unlikely to achieve the returns they want. Ultimately, if you take a human-centred approach, you are much more likely to achieve measurable value.