Malaysia's labour market has rarely looked healthier, on paper. In the first quarter of 2025 (1Q2025), the unemployment rate stood at 3.1%, with just 526,300 unemployed persons in a labour force of 17.23 million. By June 2025, the rate had further declined to 3.0%. But these rates mask troubling structural problems in the labour market. The youth unemployment (aged 15-24) rate stood at 10.2% in 2025, about three times the national rate. Over a third (35.5%) of employed graduates are underemployed, working in semi/low-skilled jobs. 

Technology-driven job anxiety is not new. British textile workers feared the mechanical loom technology in the late 1700s and early 1800s. Their fears were unfounded. The cotton gin eliminated hand-picking, but it created demand for textile machine operators. However, artificial intelligence (AI) may be different in at least two critical ways. First, its speed. ChatGPT reached 100 million users in just two months, a pace that took the telephone 75 years to reach. Second, its breadth. Unlike the steam engine, which replaced muscle, or the computer, which replaced clerical routine, AI is simultaneously threatening both white- and blue-collar jobs, e.g. the radiologist reading an X-ray, the lawyer drafting a contract, the truck driver navigating a route, and the restaurant waiter taking food orders. 

The evidence is catching up with the anxiety. Renowned labour economists Daron Acemoglu and Pascual Restrepo's important 2022 research finds that 50% to 70% of changes in wage structure over four decades in the US are explained by task displacement from automation. In an International Monetary Fund’s 2024 research, it estimates that about 40% of employment in emerging markets like Malaysia is exposed to AI disruption. A 2020 McKinsey analysis, conducted before the AI explosion, estimated that automation could displace the equivalent of 4.5 million workers in Malaysia by 2030. In 2024, a TalentCorp national workforce study in Malaysia identified 620,000 jobs at high risk of replacement, led by wholesale and retail trade (245,000 jobs), food manufacturing services (203,000), and global business services (89,000).

So, what does adaptation actually mean in practice? It is not telling everyone to learn how to code. A factory worker whose assembly line is now monitored by AI needs to learn to read the AI's warning alerts and decide when to call for repairs. An accountant no longer needs to spend hours reconciling spreadsheets, because AI does that faster. Now, an accountant can devote their time advising clients on a complex tax structure. For would-be university graduates, a future-proof career strategy is to build real depth in their discipline, while developing sufficient proficiency with AI tools.

How should the gaps be closed? Employers need to not only invest in AI to cut costs but also to invest equally in helping their workers adapt. Trade unions need to negotiate employment contracts that include the right to retraining when AI adoption changes a worker's role, and not just the right to a redundancy severance payout. The government needs to redesign schools and universities’ curriculum, and financial safety nets which allow AI-displaced workers to retrain without falling into poverty.

Professor Dr Soon Jan Jan is a senior research fellow at the Economic and Financial Policy Institute (ECOFI), Universiti Utara Malaysia (UUM).