Why AI investment is not reaching the bottom line, and why the answer is human rather than technical.
The technology works. That is not in dispute.
Large language models can draft contracts, write code and summarise research faster than any team. Most organisations have invested. Many employees are using the tools every day. Ask individuals and they will usually tell you AI makes them more productive.
So why are the gains not showing up in the results?
The Gap Between Activity and Outcome
Gallup's State of the Global Workplace 2026 report brings together some uncomfortable figures. An MIT study found that despite around $40 billion of enterprise investment, 95% of organisations had seen zero measurable impact on profits. A National Bureau of Economic Research survey of nearly 6,000 executives in the US, UK, Germany and Australia found that 89% had seen no impact of AI on their company's labour productivity over the past three years.
Gallup's own data shows the same pattern from the employee side. Among US workers in organisations using AI, 65% say it has improved their personal productivity. Only 12% strongly agree that it has transformed how work gets done in their organisation.
Individual gains are real. Organisational gains are elusive. Something is getting lost between the two.
“Even the most sophisticated technology cannot overcome an indifferent team leader.”
The Missing Link Is the Manager
Gallup's answer is one the corporate world has largely overlooked. In organisations investing in AI, the strongest predictor of employee adoption, aside from technical integration, is whether their direct manager actively champions it.
The difference is dramatic. Employees who strongly agree that their manager actively supports their team's use of AI are 8.7 times as likely to strongly agree that AI has transformed how work gets done, and 7.4 times as likely to say AI gives them more opportunities to do what they do best.
Yet fewer than a third of US employees in organisations implementing AI strongly agree their manager actively supports their team's use of it. In Germany, the figure was just 21%.
This is not new. A decade ago, researchers at Stanford, Harvard Business School and MIT found that differences in management practices accounted for around 30% of the variation in total factor productivity, the standard measure of how much value technology adds. Great technology has always needed great management to pay off. The financial stakes are simply far higher now.
The line manager is vital in enabling AI/Human Augmentation. Photo Getty Images Unsplash
Feeling Faster Is Not the Same as Being Faster
There is another reason the numbers are not adding up. People are not always good judges of their own productivity.
A 2025 randomised controlled trial, reviewed by The Oxford Review, studied experienced software developers working on real tasks. Before using AI tools, they predicted a 24% speed-up. Afterwards, they estimated they had been 20% faster. In reality, they were 19% slower.
That is one study, in one profession, and the tools are improving quickly. The lesson is broader, however. If people believe AI is saving them time when it is not, organisations will make decisions based on gains that do not exist. Somebody needs to be asking the critical questions, and that is a leadership behaviour, not a software feature.
Who Leads, the Human or the AI?
A further 2025 study reviewed by The Oxford Review looked at how the structure of human-AI working affects behaviour. When employees led and AI assisted, workplace procrastination decreased. When AI led and employees assisted, procrastination increased. Boredom explained much of the difference.
That finding matters for every organisation designing AI into its workflows. Handing the lead to the technology may look efficient on paper. In practice, it can leave people disengaged from work they no longer feel they own, which is the opposite of what the investment was meant to achieve.
This is why I describe the future of work as human and AI augmentation, not replacement. The question is not what AI can do. It is how people and AI work together so that people stay engaged, accountable and in charge of the judgement calls.
Disengagement Will Erode the Gains
Gallup describes engagement as a measure of readiness for change, and AI is a major change. With around 80% of the global workforce not engaged, and manager engagement down to 22%, most organisations are trying to adopt a transformational technology with a workforce that is only partly committed to the work.
Gallup puts it plainly: in the age of AI, productivity gains will depend partly on how effectively individual workers use the tools. Disengagement will erode those gains.
Uncertainty about AI adds to the challenge. Gallup found 23% of US employees in organisations that have implemented AI think it is likely their job will be eliminated within five years. Earlier research reviewed by The Oxford Review found that uncertainty about the impact of AI had a significantly negative effect on the performance of employees with lower digital literacy. Fear does not make people better adopters.
The recent sell-off in US data centre stocks after the midterm elections is another reminder that the AI story is still being written. Organisations cannot control the regulation or the markets. They can control how well their people are led through the change.
The Human Skills That Matter More, Not Less
If AI is taking on more of the production of work, the human contribution moves towards the things AI does least well. Four stand out:
• Critical thinking, to judge whether the output is accurate, complete and appropriate for the context.
• Decision-making, to weigh trade-offs, take responsibility and commit to a course of action.
• Relationship building, because trust, influence and collaboration remain stubbornly human.
• Leading change, because people adopt new ways of working through other people, not through software.
These are not new skills. They are the core of good leadership. What has changed is their value. When everyone has access to the same tools, the difference between organisations increasingly comes down to the quality of the human judgement applied to them.
A Quick AI Health Check for Your Leadership Team
Before investing further, it is worth asking your senior team five honest questions:
1. Can we name a business outcome, not just an activity, that AI has measurably improved?
2. Do our managers actively support their teams' use of AI, or simply allow it?
3. In our AI-enabled processes, who leads: the person or the technology?
4. When did someone last challenge an AI-generated output, and what happened?
5. Have we talked openly with our people about how AI will affect their roles?
If most of the answers are uncertain, the next investment probably belongs in leadership rather than licences.
This is particularly relevant for UK organisations. With Gallup's UK data putting engagement at around 10%, and budgets under pressure from higher interest rates, many businesses are hoping AI will deliver the productivity gains that people have not. The evidence suggests it is the other way round. AI amplifies the way people already work. Engaged, well-led teams tend to use it to get better. Disengaged teams tend to use it to do the minimum faster.
What Leaders Need to Do Differently
The behaviours that turn AI activity into AI outcomes are human ones:
• Champion it visibly. Use the tools yourself, talk about what works and what does not, and make experimentation safe.
• Keep people in the lead. Design work so that people direct the AI, rather than the other way round.
• Ask the critical questions. Is this actually faster? Is the output right? What are we no longer checking?
• Address the fear directly. Be honest about what AI will and will not change in your team's roles.
• Measure outcomes, not usage. Logins and prompts are activity. Productivity, quality and customer results are outcomes.
Where This Sits on The Resilience Wheel
AI adoption is an Adaptability challenge first. It also depends on Confidence, because people who trust their own judgement are better at knowing when to trust the machine, and on Purpose, because people engage with change when they understand why it matters and what part they play in it.
The organisations that spring forward with AI will not be the ones with the most licences. They will be the ones whose leaders helped their people learn, adapt and stay in charge of the work.
For L&D Professionals
If your organisation's AI investment is not yet showing in the numbers, look at manager capability before you look at the technology. Gallup's data points to manager support as the strongest human driver of adoption, yet most AI rollouts train people on tools rather than training managers to lead the change. A short programme helping managers champion AI, address their teams' concerns and keep people in the lead is likely to do more for your return on investment than another round of tool training.
Is your AI investment reaching the numbers?
Find out whether your leaders and teams are ready to lead the human side of AI.
Resilient Leadership Diagnostic:
theresiliencecoach.co.uk/resilient-leadership-diagnostic
Resilient Teams Diagnostic:
theresiliencecoach.co.uk/resilient-teams-diagnostic
Book an introductory call: