The rise of Artificial Intelligence lights and shadows

Anyone who has used Artificial Intelligence – even just to draft an email – will have realised that this technology is far more than an abstraction or a financial bet. AI is a tangible force, rapidly embedding itself in the real economy and reshaping how hundreds of millions of people live and work.

In this document, a year ago, we were asking ourselves if and when AI would begin to change the way we work. Twelve months later, as we write, there is evidence that many companies are already recording significant efficiency gains thanks to AI, while new applications continue to emerge every day.

Despite this evidence, in recent months investors have largely focused on big tech earnings, looking for confirmation that investments in AI have been worthwhile and that today’s elevated market valuations are justified. While this is useful for assessing short-term market performance, we believe it risks distracting from the real underlying dynamic. AI is not a gadget that will simply boost earnings over the next 12 months it is the potential trigger for an industrial transformation that could prove even more profound than the digital revolution of the past 40 years.

From an investment perspective, we usually care about the impact on profit growth, relative to starting expectations. However, we believe it is also important to step back from the noise surrounding market performance. The central question is not if or when AI will reshape production processes, organisational structures, and corporate profitability, but how deep these changes will be. We see little doubt that AI will catalyse growth and innovation; what remains to be understood is the magnitude of the wave and whether the global economy will absorb the impact without significant shocks.

What makes the AI revolution unique

AI’s revolutionary potential lies in the fact that it is a new general-purpose technology, like electricity or the internet. This means it can be applied to an extremely wide range of professional – and increasingly, personal – activities.

Compared to past breakthrough technologies, however, two crucial differences emerge. The first, which makes AI historically unique, is that it can replace cognitive and creative labour on a large scale – tasks that have traditionally been the exclusive domain of humans. The second is the speed of adoption despite the anxiety generated by quarterly earnings readings, AI uptake is advancing at an accelerated pace, supported by unprecedented investment. If the internet spread over decades, AI is penetrating industries and job functions at a much faster rate.

This uniqueness, however, introduces an element of uncertainty into a narrative that would otherwise appear unambiguously optimistic. It remains to be seen whether the economy can absorb this technology with the same adaptability shown in the past or whether the speed and depth of the change will impose adjustment costs higher than expected.

How far will AI transform the labour market?

To assess the scale of the AI revolution, the first parameter to consider is the improvement it brings to production processes. Early signals already show that AI is reshaping business operations at every level. Although evidence is still somehow contested, preliminary studies confirm that AI has the potential to generate significant productivity gains.

According to researchers at the London School of Economics (based on data from 3,000 workers and 240 managers), integrating AI into work processes could save the equivalent of a full working day per week (7.5 hours). This corresponds to a productivity boost of around 20%, quantified at roughly £13,000 per worker based on the average salary of respondents. These findings align with corporate expectations Infosys reports that companies anticipate an average 15% productivity increase from AI projects, with peaks up to 45%.

These early efficiency gains are typical of major technological revolutions. Yet it’s important to keep in mind that these numbers likely underestimate AI’s full impact. Many workers are already using AI informally, and only a small share of companies have integrated it into their core processes. According to several studies (McKinsey, Infosys), only about 1% of firms have fully embedded AI into core operations, and only 2% are ready to do so. Most remain in the pilot phase or use AI only marginally. This is crucial it suggests that most productivity gains have yet to materialise.

An increase in productivity per hour worked in these areas could – within a few years – deliver results comparable to those that would otherwise require an entire economic cycle. Estimates do vary, but for example, the OBR thinks that its realistic best-case is to end up with a productivity boost of 0.8pc annually in the UK. That would be more productivity growth than Britain has averaged since the financial crisis (0.5pc) so total productivity growth could hit 2 per cent – territory not seen since the early 2000s. For context the introduction of computers had almost no visible impact on productivity for decades, and the effects of electrification only became evident 40 years after its adoption. These observations remind us that the relationship between new technologies and productivity is often nonlinear and difficult to predict (the “productivity paradox”).

Even allowing for caution, a productivity boost smaller than current estimates would still result in substantial economic benefits higher output with the same resources, improved corporate margins, support for real wages, and a structural increase in the economy’s growth potential – translating into positive outcomes for investments. After a decade of weak growth, AI may well become the catalyst for a new cycle.

However, these positive macroeconomic effects on growth and employment will likely emerge only in a few years. The primary reason is the low formal adoption rate macro effects typically appear only once diffusion surpasses certain critical thresholds, as happened with the internet between 1995 and 2005. Until adoption gains momentum, the impact on GDP will remain muted. It is very likely that AI will become a decisive engine for global growth, but we may only start to see this trend later in the decade – barring possible downward pressure on prices and employment. Some estimates predict an additional 7% boost to economic growth in the coming years.

Expansion into consumer sectors

A step-change in adoption will occur when AI becomes integrated into mass-market consumer products, giving rise to entirely new product categories and consumption habits. Today, consumer-facing AI is still at an early stage there is no defining product – no “iPhone moment” – that can trigger mass uptake.

However, early signs are emerging in existing devices. AI integration in smartphones, wearables, and smart glasses is only just beginning. Major software and hardware makers are testing AI-agent functions embedded directly in operating systems (generative video editing, contextual personal assistants, real-time productivity tools). Still, none of these features has yet become a true sales driver.

This is likely to change. According to the World Economic Forum, AI could unlock $1.2 trillion in value across consumer sectors by 2038 – almost equivalent to the entire global luxury industry. The report suggests that AI’s spread across consumer products will act as a powerful growth engine, revitalising mature categories such as retail, entertainment, digital health, smart home devices, and food & beverage.

Labour-market effects

While the growth effects may emerge over the next decade, the labour-market impact is already visible. For example, in the UK the number of job vacancies has fallen from 1.3 million in May 2022 to around 0.7 million in May 2025. Studies suggest that AI adoption may account for part of this decline. Research from King’s College London shows that firms most exposed to AI have already reduced employment by roughly 4.5% compared to others – a first measurable impact of AI on the labour market.

Although still marginal relative to the scale of the transition, this raises an important question once AI adoption reaches maturity, will the economy be strong enough to absorb the shift? Historical fears about the “end of work” have accompanied every major technological innovationv–vand have almost always proved exaggerated. However, AI represents an unprecedented leap in productivity, and this is arguably the most significant risk factor that could challenge the anticipated AI-driven economic boom.

Goldman Sachs estimates that generative AI could automate up to one-quarter of tasks in the US and Europe, potentially affecting up to 7% of the workforce in the absence of new job creation. These numbers should not be underestimated during the 2008-09 financial crisis, US unemployment peaked at 10%. Reallocating labour on such a scale – even if ultimately absorbed – could generate political and social tensions and pose new challenges for policymakers. This process will also take time, while job displacements might happen quicker. 

Public policy will be essential to ease the transition, supporting welfare systems and workforce training. The World Economic Forum estimates that 59% of workers will require reskilling or upskilling by 2030. Many companies are already moving in this direction 77% expect to invest in skill development, even while acknowledging that some roles will shrink. Governments and firms will need to coordinate on education programmes, professional training, and social-safety measures. How quickly policymakers act will increasingly become a key variable.

In the meantime, the economy will likely reorganise itself gradually. The prevailing view is that AI will transform and reallocate jobs rather than eliminate them. A global survey shows that 86% of companies expect AI to transform their industry by 2030 – and crucially, they also expect new roles to be created. Latest projections from the WEF even suggest a net positive when considering the overall job market 170 million new jobs could be created by 2030, compared to about 92 million eliminated–a net gain of 78 million.

Predicting the exact scale of these effects is difficult, but we believe this dynamic between rising productivity and labour rotation will become the defining economic process of the next decade. There will be disruptions and difficult transitions, but over the long term, AI appears as a powerful engine of progress–and a clear signal for investors to remain invested with a long-term perspective.

Progress, especially when it is as disruptive as it is today, needs to be managed. It is understandable that investors’ attention is currently focused on valuations and on the concern that enthusiasm around Artificial Intelligence may have inflated the market. However, several factors clearly distinguish the current phase from similar historical episodes, such as the dot-com bubble.

Unlike in the past, many of the companies driving this transformation today have well-established business models, high levels of profitability and significant cash flows, enabling them to fund investment without relying on purely speculative expectations. Moreover, as discussed in the article, AI adoption is still in its early stages, suggesting that the central issue is not excessive maturity, but rather how to manage a profound technological transition.

Looking at history, around 60% of US workers are employed in occupations that did not exist in 1940, and more than 85% of employment growth since then has come from new roles created by technological progress. For this reason, we believe investors should welcome the start of the AI era with confidence. The most effective way to take part in this new wave of progress is to remain invested in the markets, with careful attention to risk control and diversification.

This article is part of our new Strategic Asset Allocation, which sets out our long-term view on markets and informs the strategic positioning of the portfolios managed by our team on your behalf.