More than a decade ago economist Carl Benedikt Frey and scientist Michael Osborne made a shocking claim that technological disruption would threaten 47% of jobs in the United States, provoking a widespread fear of automation displacing humans and destroying jobs.

This intensified as artificial intelligence (AI) began to integrate rapidly into the global economy, but new economic thinking suggests something different.

Economic theory and empirical evidence show that AI replaces routine and replicable activities and simultaneously catalyses the creation of entirely new, higher-value occupations. The net employment effect of technological innovation is mostly positive.

In addition, “technology-resistant” sectors characterised by high human contact, emotional intelligence and complex personal interaction, continue to be a substantial component of employment.

Jobs in healthcare, advanced education and specialised hospitality rely on intrinsic human capabilities that cannot be completely automated. So, AI is poised to restructure the workplace rather than precipitate a jobless dystopia.

The main economic threat posed by AI is far more insidious, targeting labour remuneration rather than the number of employees. AI may not take your job but it will almost certainly take your salary.

There is a relationship between technological adoption, productivity and wealth distribution. Economists have long observed this through the “Solow paradox”, where digital transformation is visible across industries but does not immediately impact aggregate productivity statistics — you can see the computer age everywhere except in the productivity statistics.

This lag is sometimes called the productivity “J-curve”. When a general-purpose technology (GPT) like AI is first introduced productivity growth typically stagnates or declines.

This is because firms commit substantial capital, time and organisational restructuring to master and integrate the new infrastructure.

Productivity improves only after this costly adjustment phase.

This makes the workforce vulnerable. While standard economic theory suggests that real wages are tied to productivity, if productivity growth stagnates during the initial downward phase of the J-curve, adopting AI will not translate into higher wages.

Also, if AI disruptions force labour market restructuring and worker displacement, downward pressure on wages will intensify and historical precedents suggest that wage recovery can be a protracted, even multi-decade process.

In the context of developing and transition economies like Malaysia, this structural vulnerability is severely compounded by rigid institutions.

Malaysia has a low compensation of employees (CE) ratio, which consistently hovers around 32% to 33% of gross domestic product (GDP). This compares to advanced economies, where the labour share of income typically ranges between 40% and 50%.

The implication of a depressed CE ratio is serious. When productivity gains finally materialise at the top of the J-curve, the financial fruits of that progress are distributed unevenly.

Under this framework, most AI-driven productivity gains will accrue directly to employers rather than employees. While businesses will experience higher profit margins and operational efficiency, the labour market structure denies employees the collective bargaining power necessary to capture an equitable share of that extra value-added.

When businesses control all the leverage from AI, individual workers face an ultimatum to accept lower pay from a new scope of work or be replaced by a more compliant group of workers including underemployed graduates and women entering the workforce in part-time jobs.

The macroeconomic future under an AI regime is not one of vacant offices but rather one of fully staffed enterprises where workers generate exponentially higher output for stagnant or diminishing real wages.

To deal with this systematic suppression of salaries, policy must pivot away from futile attempts to restrict technological adoption and focus on reforming and freeing the labour market to ensure a more equitable distribution of the digital dividend.

Otherwise, there will be more jobs with lower wages.

The views expressed are those of the writer and do not necessarily reflect those of FMT.