Let’s paint an image. An organisation decided to implement AI use in their workflow. It hired experienced consultants and invested in a modern AI platform. The system was tested, approved, and deployed on schedule.
They did everything right, and on paper, the project was a success. Then six months later, the staff went back to doing things the old way. Some kept their own spreadsheets.
Others continued using WhatsApp messages instead of the new workflow. A few used the AI system only when someone from management was watching.
The technology hadn’t failed, the implementation had. And the reason had very little to do with software and everything to do with people.
In Part 5 of this series, we looked at the practical first steps Nigerian public institutions can take to begin their AI journey.
This article looks at the step most organisations skip entirely, bringing the people along. This is one of the biggest misconceptions surrounding AI adoption.
When an implementation struggles, organisations often assume they chose the wrong technology.
Sometimes that’s true. More often, the technology worked exactly as promised. What failed was trust.
The pattern we’ve seen before
Whether it’s an AI system, a new records platform, or a digital workflow, failed implementations tend to follow a familiar pattern.
Leadership identifies a need for change. An IT team or external vendor is brought in. The system is configured, tested, and launched. Staff are invited to a training session, given new login details, and expected to adapt.
Then something interesting happens.
People don’t openly reject the system. They simply stop using it.
Not because they’re lazy. Not because they’re resistant to innovation. But because the new system doesn’t fit the way they actually work, or because nobody took the time to explain why the change mattered in the first place.
Technology is introduced. People are expected to catch up. When they don’t, the project is labelled a failure.
The fear nobody talks honestly about
Whenever AI enters the workplace, one question appears almost immediately.
“Is this going to replace my job?”
It is a deeply human question. And yet it is often met with a rehearsed response.
“Don’t worry. AI won’t replace people. It will only help you work more efficiently.”
Sometimes that’s true. Sometimes it isn’t, and organisations should be honest about that.
In roles built almost entirely around repetitive data entry, document sorting, or processing high volumes of identical requests, AI can and does reduce the number of people needed for that specific task.
Pretending otherwise does not protect employees. It simply delays a conversation that will happen anyway.
The problem isn’t that organisations talk about the benefits of AI. The problem is that they sometimes avoid having an honest conversation about what change actually means.
In many Nigerian workplaces, employment is more than a salary. One income may support parents, siblings, children, or extended family members.
Losing a job doesn’t affect one person, it can affect an entire household.
Against that reality, fear is not irrational. It’s understandable.
If an administrator in a public university hears that AI can process documents faster, they may wonder whether fewer administrators will eventually be needed.
If a civil servant hears about automated report generation, they may question how their role will change.
Ignoring those concerns doesn’t make them disappear. It simply drives them into private conversations where uncertainty grows.
The most effective leaders don’t pretend these fears don’t exist. They acknowledge them. They explain what is changing, what isn’t, and where human expertise remains essential.
Even when there are difficult conversations to be had, honesty builds far more trust than vague reassurance ever will.
Consultation is not the same as participation
Many organisations believe they involve staff in major technology projects. What they actually do is announce a decision.
A town hall meeting is organised. A presentation is delivered. Questions are invited. By that point, however, the software has already been purchased, the implementation timeline agreed, and the decisions made.
That isn’t participation. It’s communication. Real involvement starts much earlier. It begins by asking the people closest to the work what slows them down every day.
- Where do mistakes happen?
- Which tasks consume the most time?
- What would make their jobs easier?
Those questions do something important. They shift staff from feeling like change is happening to them to feeling like they are helping shape it.
When employees help identify the problem, they are far more likely to believe in the solution.
Design for the people doing the work
One of the easiest mistakes to make is designing systems around organisational charts instead of real workflows.
A process may look efficient in a meeting room. It often looks very different at a university admissions desk, a hospital records office, or a government service counter.
The people doing these jobs know where delays occur. They know which forms are confusing. They know where citizens become frustrated.
They know which shortcuts exist, and why they exist. If those people are excluded from the design process, AI may solve the wrong problem. The result is a technically impressive system that makes everyday work harder rather than easier.
The best technology often feels almost invisible, and in a Nigerian public institution context, that might mean an AI tool that slots into an existing WhatsApp-based workflow rather than replacing it entirely, or a document processing system that mirrors the manual steps staff already know rather than introducing an entirely new logic.
When people do not have to think about the tool, they can focus on the work.
Trust cannot be rushed
Trust isn’t installed with software. It develops over time. People need to understand what an AI system does. Just as importantly, they need to understand what it does not do.
- Can it make mistakes? Yes.
- Who checks its outputs?
- What happens if it gets something wrong?
- Who is accountable?
These questions shouldn’t be treated as obstacles. They are the foundation of trust.
Accountability should sit with a clearly named person or team, not with the software vendor, and not vaguely with “the organisation.”
In practice, this means designating a responsible owner for every AI system: someone whose job includes monitoring outputs, fielding concerns from staff, and escalating problems when they arise.
Without that named accountability, errors go unaddressed and trust erodes quietly.
When organisations are transparent about AI’s strengths and limitations, people become more confident using it responsibly.
Ironically, admitting that a system isn’t perfect often increases confidence far more than pretending it is. Trust grows through consistency. Not marketing.
The people outside the organisation matter too
Public institutions don’t exist only for their employees. They exist to serve people.
A student applying for admission. A patient seeking healthcare. A citizen renewing a licence. A business owner applying for a permit.
These people may never see the AI system working behind the scenes. But they experience its decisions. If an application is delayed because an automated process failed, they deserve to know what happened.
If AI helps prioritise cases or process requests, there should still be clear ways to ask questions, seek clarification, or challenge a decision when necessary.
Efficiency should never come at the expense of accountability. Public trust depends on more than delivering services quickly. It depends on delivering them fairly and transparently.
Success looks quieter than people expect
When AI adoption succeeds, it rarely makes headlines. Consider a state government ministry that introduced an AI tool to help process citizen enquiries.
No launch event. No press release. Within three months, response times had halved and staff were spending less time on routine queries and more time resolving complex cases that genuinely needed human judgment.
The technology became part of how the office worked. Nobody talked about it anymore, because it was simply working.
That is often the strongest sign that implementation has succeeded. Staff spend less time on repetitive administration.
Citizens receive quicker responses. Students navigate processes more easily. Managers have better information for making decisions.
Most importantly, people stop talking about the technology. It simply becomes part of how work gets done. The focus shifts from the software to the people it enables.
Closing thought
AI projects rarely fail because the algorithm wasn’t sophisticated enough. More often, they fail because the people expected to use the technology were treated as an afterthought.
Technology can automate tasks. It cannot automate trust. That has to be earned, through honest conversations, meaningful involvement, thoughtful design, and the willingness to listen before deciding.
The most successful AI systems are not the ones with the most advanced features.
They are the ones people choose to use because they understand them, trust them, and see that they genuinely make their work, and the lives of the people they serve, a little better.
Because in the end, AI is never just about machines. It’s about people. And people should never be the last thing we think about.
