Malaysia's AI moment: Dynamism without displacement

LETTER: A 2026 survey of nearly 9,000 executives across seven economies found that AI adoption raised annual labour productivity growth by just 0.2 per cent over the past three years, though executives expect a 0.6 per cent boost going forward.

Malaysia's National AI Action Plan 2026–2030 targets 0.8 to 1.2 percentage points of incremental GDP growth annually — roughly RM13–RM20 billion — along with 300,000 to 500,000 new AI-related jobs and top-20 global AI readiness by 2028.

It is also estimated that approximately 685,000 Malaysian workers could be significantly impacted within three to five years without accelerated reskilling.

One research suggests 45 per cent of the workforce — about 6.7 million workers — have at least 40 per cent of job tasks potentially substitutable by existing generative AI technologies.

The central policy tension is, AI adoption without workforce transition will create dislocation; workforce transition without AI adoption will leave Malaysia uncompetitive.

The framework must rest on two pillars. The first is AI for economic dynamism — raising the ceiling for the economy. Three channels matter most.

First, MSME (Micro, Small, and Medium Enterprises) productivity. For Malaysia, where MSMEs form the backbone of employment, the priority is diffusion, not invention: getting proven tools into the hands of 4,000 MSMEs and ensuring they achieve measurable efficiency gains.

Globally, 76 per cent of small business owners use AI, with 93 per cent reporting positive impact.

But only 14 per cent have fully integrated AI into core operations. The barrier is not willingness — it is technical expertise and implementation support.

Second, public sector efficiency. The Bank of England's use of AI to extract signals from corporate reports and real-time financial data offers a model for central banks and statistical agencies.

Third, data centre and digital infrastructure spillovers.

Malaysia's data centre investments are already lifting construction activity. The strategic question is how to capture more of the value chain — not just hosting, but AI-related services and domain-specific applications.

The second pillar is AI for economic stability — reducing the severity of shocks and strengthening recovery capacity.

A credible strategy must address risks directly. US AI spending is highly concentrated — just two per cent of firms account for over 50 per cent of total AI spending.

If productivity gains do not materialise at scale, a correction could be sharp. Malaysia's technology-linked sectors are not immune.

AI's net employment effect may be small, but the churn is large: over 40 per cent of firms anticipate positive or negative employment effects that roughly offset only in aggregate.

Behind the numbers are real people in routine clerical roles who may not transition smoothly.

The choice is not between AI adoption and stability. It is between managed transition and unmanaged disruption.

AI can raise productivity, improve public services, and strengthen risk detection — but only if institutions, incentives, and support systems ensure that gains are real and costs are shared.

The task is to turn that principle into measurable outcomes and to ensure that our AI moment delivers dynamism without displacement.