For many organisations, AI still feels like something that belongs on the strategy agenda. Leadership teams are discussing which tools to invest in, where automation might make a difference and what the longer-term impact on their workforce could be.

Meanwhile, employees are getting on with it.

Across almost every business function, people are experimenting with generative AI to draft communications, analyse information, summarise documents, generate ideas and speed up everyday tasks. Recent research from Skills England suggests that this is becoming increasingly commonplace, with AI now being used daily in a significant proportion of workplaces.

That creates an interesting challenge for employers, because AI adoption isn’t necessarily happening in the neat, structured way that organisations might expect. In many cases, it is developing from the bottom up, with employees learning through experimentation, colleagues and the tools themselves.

AI adoption is as much a people challenge as a technology challenge

It is easy to approach AI transformation as an IT project. An organisation chooses a platform, gives employees access, introduces a policy and perhaps runs some introductory training.

Technology is obviously an important part of the picture, but access to AI doesn’t automatically create organisational capability.

To use AI effectively, businesses need people who can recognise where it genuinely adds value, understand the processes they are trying to improve and assess whether a new way of working has actually produced a better result. They also need employees who are confident enough to question an AI-generated answer rather than simply accepting it.

At leadership level, the requirements are different again. Senior teams need enough understanding to evaluate opportunities, ask sensible questions about risk, establish appropriate governance and decide where investment will make a meaningful difference.

Interestingly, this combination of technical understanding and human capability is reflected in LinkedIn’s 2026 Skills on the Rise research. AI-related capabilities feature prominently, but so do skills such as collaboration, communication, stakeholder management and leadership.

The bigger risk may be fragmented adoption

One of the less discussed challenges of AI is what happens when employees embrace it successfully, but the organisation doesn’t capture what they have learned.

Over time, businesses can develop pockets of highly effective AI use without fundamentally improving how the wider organisation operates. Individual productivity increases, but the knowledge behind those improvements stays with individual employees or teams.

There is also the possibility that several people are independently trying to solve exactly the same problem.

For leaders, there is an opportunity here to understand what employees are already doing, identify what is working and create ways for useful practice to spread. In some organisations, the starting point for an AI strategy may therefore be much simpler than expected: find out what is already happening.

Start with the business problem, not the AI tool

There is another temptation that comes with any rapidly developing technology: starting with the tool and then looking for somewhere to use it.

Once those problems are visible, AI and automation become possible solutions rather than the objective themselves.

This matters because the people closest to those problems are often the people best placed to identify the opportunity. A finance team understands the repetitive work within finance. Operations teams know where bottlenecks occur. HR teams understand their administrative burden, while sales teams know which activities consume time without contributing much value.

Organisations don’t necessarily need every employee to become an AI specialist. There is enormous value in helping people who already understand the business to develop enough AI and automation capability to improve it.

Leaders and practitioners need different skills

One of the areas employers may need to think about carefully is the difference between using AI and leading an organisation that uses AI.

An employee who is redesigning a process needs practical skills. They need to understand the existing workflow, identify inefficiencies, explore appropriate technologies, consider the risks and measure whether the change has delivered the expected result.

A senior leader doesn’t necessarily need to know how to build that solution themselves, but they do need enough understanding to evaluate the proposal. What problem are we solving? What information will the technology access? What are the risks? How will people be affected? What happens when the technology gets something wrong? How will we know whether the investment has worked?

Without that understanding, organisations can end up at either extreme. They may have plenty of enthusiasm and experimentation without sufficient governance, or they may have an ambitious AI strategy that never quite translates into changes in day-to-day work.

AI literacy is becoming part of everyday workforce development

Perhaps the most important change is that AI literacy is starting to look less like a specialist digital skill and more like a general workplace capability.

That doesn’t mean everybody needs to understand machine learning models or become an AI engineer. For most employees, useful AI literacy is much more practical. It means understanding what the technology can and cannot do, recognising where it could improve their work, knowing how to use it responsibly and being able to assess the quality of its output.

Human judgement remains an important part of that equation. Being able to ask a good question, recognise unreliable information, understand context and communicate effectively doesn’t become less valuable when AI enters the workplace. If anything, those skills become more important.

Turning experimentation into organisational capability

There isn’t a single AI roadmap that will suit every organisation, and trying to create one before understanding what is happening inside the business may be counterproductive.

A useful starting point is to speak to employees and find out where AI is already being used. Leaders can then look at which business problems could realistically benefit from AI or automation, who understands those processes well enough to improve them and where additional skills would help.

It is equally important to decide how success will be measured. The number of AI licences purchased tells an organisation very little about whether AI adoption is working. Time saved, errors reduced, capacity created, faster customer response times, improved employee experience or measurable process improvements are much more meaningful.

Time saved	Errors reduced	Capacity created
Faster responses	Better experience	Process improvement

There will inevitably be experimentation along the way. Some ideas will work brilliantly, while others will prove that the existing process was better after all. Creating the capability to test, evaluate and learn from both outcomes is arguably more valuable than trying to predict every successful use case in advance.

The most important AI investment might be your people

The AI tools businesses use will continue to change quickly. The platform dominating today’s conversation may have completely different capabilities in two years’ time, and entirely new tools will inevitably appear.

That makes focusing solely on technology a difficult long-term strategy.

An employee who understands their organisation, can analyse a process, identify opportunities for improvement, evaluate technology critically and implement change has developed a capability that can survive those changes. Similarly, a leader who understands enough about AI to make informed decisions will be better equipped to respond as the technology develops.