AI adoption surges, but the productivity dividend remains elusive

Artificial intelligence has rapidly moved from experimentation to everyday business practice. Employees across industries are using AI tools to draft reports, analyse data, automate administrative tasks and accelerate decision-making. Yet despite growing confidence in the technology, many organisations are struggling to translate these gains into improved earnings.

According to McKinsey’s State of AI research, 80 per cent of respondents say AI has improved their productivity. However, only 37 per cent of organisations report any direct impact on earnings, a figure that has remained largely unchanged over the past year. The findings point to a widening gap between individual efficiency gains and organisation-wide business performance.

The disparity raises an important question. If employees are working more efficiently, why are so many companies failing to realise a meaningful financial return?

Productivity alone does not create value

One explanation is that productivity improvements do not automatically translate into profitability. Employees may complete tasks faster, spend less time on repetitive activities or generate content more efficiently, but those gains only create economic value when organisations redesign workflows, increase output, improve customer service or reduce operating costs.

Many businesses remain in the early stages of AI deployment. Staff use generative AI to summarise documents, create presentations, draft correspondence and support research activities. While these applications save time, the benefits often remain isolated at the individual level and are difficult to measure consistently across an enterprise.

Moreover, organisations frequently attempt to layer AI onto existing processes rather than fundamentally rethinking how work is performed. In such cases, productivity gains can become diluted. A task completed twice as quickly does not necessarily increase revenue or reduce expenditure if surrounding systems and processes remain unchanged.

This pattern echoes earlier waves of technological innovation. Research from the Organisation for Economic Co-operation and Development (OECD) has highlighted how the benefits of major technologies often emerge gradually, as organisations adapt their operating models to take advantage of new capabilities. Similarly, the World Economic Forum has noted that the greatest value from AI is typically realised when technology, processes and culture evolve together.

The emergence of the AI-enabled workforce

What is clear is that AI has become an increasingly normal part of professional life. Rather than being viewed as a futuristic technology, generative AI is rapidly becoming another productivity tool in the modern workplace.

Knowledge workers are using large language models to generate first drafts of reports, interpret complex datasets, produce software code, translate content and automate routine administrative tasks. In many cases, AI acts as a digital assistant, enabling employees to focus their attention on activities requiring judgement, creativity and interpersonal skills.

Evidence supporting these benefits continues to grow. Research published through the National Bureau of Economic Research (NBER) has found that generative AI can significantly improve productivity in certain forms of knowledge work, particularly among less experienced employees. Such findings suggest AI may help narrow skills gaps while improving overall workforce efficiency.

Yet productivity remains a multifaceted concept. Employees may feel more productive because tasks are completed more quickly, but organisations require meaningful metrics to determine whether quality, innovation, customer outcomes and commercial performance have also improved.

The organisations reporting the strongest results from AI are often those that have progressed beyond isolated pilot projects. Successful adoption tends to share several common characteristics. First, AI initiatives are linked to high-value business objectives rather than novelty-driven experimentation. Second, AI capabilities are integrated directly into operational workflows. Third, organisations invest heavily in governance, employee training and change management.

Increasingly, competitive advantage is not determined by who has access to AI. Most organisations can now access advanced models and cloud-based AI tools. The differentiator lies in how effectively those tools are implemented.

Many businesses continue to face challenges relating to data quality, fragmented technology infrastructures, cybersecurity risks and regulatory compliance. According to guidance from the US National Institute of Standards and Technology (NIST), robust governance and risk management frameworks are essential to ensuring AI systems deliver reliable results while maintaining trust. Analysts at Gartner have similarly argued that organisational readiness often determines the success or failure of AI programmes more than the underlying technology itself. In other words, implementation is becoming more important than innovation alone.

The challenge of measuring AI’s return on investment

Another factor contributing to the gap between productivity gains and earnings impact is measurement. Traditional financial indicators often fail to capture the broader benefits associated with AI deployment. Faster decision-making, improved employee experiences, enhanced customer interactions and reduced operational friction all create value, yet these benefits may take time to influence revenue, profitability or market performance.

At the same time, deployment costs remain significant. Organisations continue to invest in infrastructure, software licences, governance frameworks, employee upskilling programmes and cybersecurity measures. Such expenditures can offset short-term gains and delay visible returns. As a result, executives face increasing pressure to demonstrate tangible outcomes while continuing to invest in long-term AI capabilities.

This complexity is reflected in analysis by the International Monetary Fund, which suggests AI has the potential to reshape productivity across large sections of the global economy. However, the pace of adoption and the distribution of benefits are likely to vary considerably between sectors, organisations and regions.

As AI adoption accelerates, governance is emerging as one of the most important determinants of success. Regulatory frameworks are evolving rapidly, particularly in Europe, where the European Union’s AI Act is introducing new obligations for organisations developing and deploying AI systems.

Businesses are therefore investing more heavily in transparency measures, risk assessments, data governance programmes and human oversight mechanisms. At the same time, the International Organization for Standardization (ISO) continues to develop standards aimed at supporting responsible and trustworthy AI deployment. Importantly, governance should not be viewed solely as a compliance requirement. Effective governance builds trust among employees, customers and stakeholders. Without that trust, adoption may stall and the potential benefits of AI could remain unrealised.

Perhaps the most important lesson from the latest findings is that AI’s economic impact remains a work in progress. The fact that four in five respondents report productivity benefits indicates that the technology is already delivering value at the individual level. Yet the static earnings figure suggests many businesses have not yet crossed the threshold from incremental efficiency improvements to genuine organisational transformation.

History offers a useful parallel. The full economic impact of technologies such as personal computing, enterprise software and cloud computing took years to emerge. Research from the Brookings Institution has shown that technological adoption often outpaces measurable economic gains, particularly during the early stages of a transformation cycle. AI may be following a similar trajectory. The technology is becoming more capable, employee confidence is growing and investment remains strong. However, the next phase of the AI journey is likely to depend less on technological breakthroughs and more on organisational change.

If businesses can successfully redesign processes, develop new operating models and embed AI into their strategic decision-making, the productivity gains currently visible at individual desks may finally begin to appear on corporate balance sheets. Until then, AI’s promise will remain compelling, but only partially fulfilled.