AI at Work: Moving From Experiment to Impact

AI is already changing the shape of the work we do and how we do it. The question for HR leaders is whether those changes happen by accident or by design.

Right now, many organisations are stuck somewhere between experimentation and impact. Tools have been approved, pilots have been launched, and teams have been encouraged to “try AI”. Yet the promised gains in productivity, creativity and innovation still feel uneven.

That’s not a surprise.

AI adoption is often treated as a technology challenge. Give people access to a set of tools and provide them with mandatory training full of use cases. Then measure adoption and move on to the next challenge.

But if the goal is real business impact, the challenge isn’t only whether people know how to use the AI tools they’ve been given. The bigger challenge is building a culture where people can experiment without fear, learn from what doesn’t work, and keep going until they find where AI can create genuine value.

Fear is stopping people from experimenting

For all the excitement around AI, many employees are still approaching it with fear.

Microsoft’s 2024 Work Trend Index found that 52% of people who use AI at work are reluctant to admit using it for their most important tasks. It also found that 53% worry that using AI on important work makes them look replaceable.

That should matter to every HR leader.

If people fear that using AI makes them look less valuable, they won’t experiment openly. They’ll hide their usage, avoid bigger opportunities, or use tools in low-risk ways that don’t change much or create any substantial business impact. In some teams, AI becomes something people play with at the edges of work, rather than something they use to rethink how work gets done.

This is where organisations can slip into what I call the Big Freeze: the pattern where teams play it safe, avoid anything that feels risky, and wait for permission at the exact moment their organisation needs them to show initiative.

People may be told that AI is important. They may know the organisation wants innovation. But if they’re uncertain about what AI means for their role, unclear on what good use looks like, or worried about being judged for getting it wrong, they’ll play it safe.

If HR leaders want AI experimentation to lead to impact, they need to reduce the threat around adoption.

Start with the impact, not the tool

A common mistake with AI is starting too close to the technology.

Which platform should we use? Which tool should we train people on? Which teams should get access first?

Those questions are, of course, important, but they are not the best starting point if your goal is long-term adoption and behaviour change.

The better question is: what impact do we want AI to help us create as an organisation?

That might be faster decision-making, better customer insight, less administrative drag, stronger knowledge sharing, improved employee experience, or more space for deep work. Each of those goals will require different behaviours, different skills and different guardrails.

Without that clarity, experimentation becomes scattered. One team uses AI to summarise meetings. Another uses it to draft content. Another tries to automate reporting. Activity increases, but the organisation still struggles to explain what value is being created.

Clarity turns experimentation from random use into purposeful learning.

For HR, this means helping leaders define the outcome before asking people to adopt the tool. What work should AI make better? What should humans be able to spend more time on? What risks need to be managed? What should not be automated, even if it can be?

Those questions are especially important because AI does not just speed up existing work. Used well, it changes the shape of work. It can shift where expertise sits, how decisions get made, how managers coach, how teams review quality, and how people spend their time.

BCG’s 2026 AI at Work research found that AI is reshaping jobs faster than companies are reshaping work. That gap is where HR has a strategic role to play.

If organisations only focus on tool adoption, they may get usage without transformation. If they focus on work redesign, they have a better chance of turning experiments into impact.

Make curiosity part of the work

Once the desired impact is clear, teams need space to explore how to get there.

This is where curiosity becomes important. Not curiosity as a vague mindset, but curiosity as a working practice.

Teams need permission to ask:

• Where could AI genuinely improve this work?
• Where might it create risk, bias or over-reliance?
• What tasks should we stop doing, not just speed up?
• What would we need to test before scaling this?
• What would make this useful for employees, not only efficient for the organisation?

These questions need time and structure. Otherwise, AI experimentation becomes an after-hours activity for the most confident or interested employees. That creates uneven adoption and risks leaving others behind.

Google’s well-known “20% time” is useful here, not because every organisation can copy it literally, but because of the principle behind it. Google has linked the approach to products including AdSense and Gmail. But the deeper lesson is not the percentage of time. It’s the licence to work on ideas beyond core responsibilities, with enough trust and space for useful things to emerge, if they do.

Senior HR leaders don’t need to promise everyone a day a week to experiment with AI. But they do need to ask where experimentation is supposed to happen.

Is it built into team rhythms? Is it protected in workload planning? Are managers expected to make space for it? Are people rewarded for useful learning, or only for immediate results?

If experimentation relies on spare time, it will always lose to urgent work.

Celebrate the learning that leads to impact

Not every AI experiment will lead to impact; that’s an important point to acknowledge.

Some will produce poor outputs. Some will reveal that the data isn’t good enough. Some will show that a workflow is too messy to automate. Some will raise ethical, legal or human concerns that need to be addressed before anything is scaled.

That doesn’t mean the experiment “failed”. It means the organisation learned something useful before spending more time, effort and resources in the wrong area.

This is where HR leaders have an important role to play. If every AI experiment is judged only by whether it creates an immediate productivity gain, people will become more cautious. They’ll share the polished success stories and hide the messy learning. They’ll avoid testing the harder, more interesting questions because the cost of getting it wrong feels too high.

That is how experimentation becomes performance theatre.

To move from experiment to impact, organisations need to protect the time and effort it takes to learn. That means recognising teams for what they discover, not just what they deliver. It means celebrating the experiment that spots a risk early, improves a workflow, creates a better review habit, challenges an unsafe assumption, or shows where human judgement matters most.

HR can help make that practical by asking leaders to:

• Define the business impact AI is meant to create before asking teams to experiment.
• Give teams clear guardrails so they know where they have permission to test and where they don’t.
• Build experimentation into the rhythm of work, rather than relying on spare time.
• Reward useful learning, including the experiments that show what not to scale.
• Share examples that include the messy middle, not only the polished result.

That is how HR moves AI out of the tool conversation and into the work conversation.

Because the organisations that benefit most from AI won’t be the ones that simply push people to use it faster. They’ll be the ones who help people experiment with enough clarity, curiosity and courage to make the technology useful.

AI at work will only move from experiment to impact when people are given the conditions to experiment well.

---

About the author:

Amale Ghalbouni is a transformation strategist, keynote speaker, executive coach and author of Experimental: The Restless Leader’s Field Guide For Building High Performing, Change-Ready Teams. For almost two decades, she has worked inside some of the world’s most risk-averse organisations, bridging the gap between what leadership wants and what teams are willing to risk. She founded The Brick Coach after years of consulting at FTSE 100 firms, with clients including Visa, Microsoft, Levi’s, Generali, Publicis Groupe, LSE, Dentsu and many more across financial services, retail, media, tech and industrials.