At risk: Tasks involving little subjective judgment are highly exposed or affected by AI use. — MUHAMAD SHAHRIL ROSLI/The Star
MARIE Abdullah, 39, was laid off as a clerk more than a year ago. The company was undergoing a retrenchment exercise, and the mother of one did not survive the redundancy chopping block.
Not all of the clerks were removed, however. Some remained because they had mastered artificial intelligence (AI) in their work – from drafting letters and proofreading documents, and even analysing presentations for meetings – making themselves essential to the company.
“So those like me who were not adept at using AI were considered redundant. It is difficult to grasp, but at the same time, a logical decision to make,” says Marie.
Her experience reflects a broader trend highlighted in a study on the effect of AI on Malaysia’s workforce by the Institute of Strategic and International Studies (Isis), the World Bank and think tank Konrad-Adenauer-Stiftung (KAS) Malaysia.
The study, which is based on the 2021 Labour Force Survey, indicates that 4.2 million Malaysian workers – or 28% of the labour force – are “highly exposed” to generative AI technologies, while another 2.5 million fall into the medium-high exposure category.
In this context, “highly exposed” refers to jobs where a significant portion of tasks could be directly affected, transformed, or even replaced by generative AI. Nearly half of the workforce, the study notes, has at least 40% of its tasks replaceable by today’s generative AI capabilities – mainly structured, screen-based, non-physical work.
Experts warn that the challenge now lies in staying relevant in a workforce rapidly reshaped by AI – one that demands workers adapt, reskill and continuously enhance their value to remain employable.
Auto-types
The study finds that jobs most exposed to generative AI typically involve predictable, repetitive or text-based activities with little need for subjective judgment. These include tasks such as typing, transcribing, proofreading, processing medical records and recording financial transactions (see “Who is most exposed?” below).
“At the same time, occupations in the highest-exposure quartile – such as software developers, financial analysts and application programmers – have an AI exposure score of 1.0, indicating that every task within these jobs could be automated by generative AI in the next three to five years.
“These occupations typically consist of structured cognitive activities, such as data analysis, coding and information processing,” the report says.
Conversely, jobs requiring significant human interaction scored as low as 0.10.
“These occupations require the interpretation of complex information, ethical sensitivity, or significant human interaction,” the study notes.
For example, the Malaysia Standard Classification of Occupations lists a judge’s job as among the least exposed to AI automation.
“The only part of a judge’s work that may involve AI is researching legal issues and preparing reports,” the study says.
“But not hearing and weighing arguments and evidence, interpreting and enforcing regulatory procedures, making rulings on admissibility of evidence, determining rights and obligations of parties, directing the jury, or pronouncing judgement.”
More retrenchments?
The findings also raise a question – has there been an uptick in retrenchments in certain roles since AI entered the workplace?
Social Security Organisation (Socso) employment services head Gayathri Vadivel says the trend is “very much” aligned with the Isis-KAS-World Bank study where the Employment Insurance System (EIS) is concerned, with clerical and administrative support services being visibly affected. However, she stresses that the causes are not solely AI-driven.
“Based on Socso’s data, we see more dominant factors such as traditional cost pressures, downsizing and resizing exercises contributing to job losses. So AI is not really ‘the’ factor when it comes to retrenchment, at least for now.”
Gayathri says for now AI is not the main factor when it comes to retrenchment.
She adds that some retrenched workers under Socso’s EIS have come from services, particularly retail, food and beverages, finance and logistics.
“Although there are cases, the numbers are not significantly high. Most retrenchments still stem from manufacturing, construction and services overall.
“Nevertheless, we do see early signs of AI exposure, especially in clerical and routine administrative roles. Many call centres and customer service positions are being replaced by chatbots or automated desk services.”
Socso recorded 56,000 job displacements in 2024, and 39,590 so far this year.
“From the 39,590, only about 3,000 came from clerical and support services. I would attribute those more to AI exposure – although not exclusively AI disruption,” Gayathri says.
A safety net
While the Isis-KAS-World Bank study offers a big-picture view of potential vulnerability across the workforce, policymakers are also looking at more immediate risks.
In August, Human Resources Minister Steven Sim said more than 620,000 jobs in Malaysia are expected to be significantly impacted within the next three to five years across 10 core sectors, according to a report by national recruitment agency TalentCorp.
That represents about 18% of 3.5 million workers in those industries – underscoring how disruption is no longer a distant possibility but an approaching reality. Sim also emphasised that the transformation will open significant opportunities for new, higher-skilled roles aligned with future needs.
TalentCorp group chief strategy officer Nazrul Aziz says it is crucial to view AI not just as a threat but as an opportunity to “ride the wave instead of becoming the victim”.
“We found that routine, transactional roles are most impacted. The three sectors expected to be hardest hit are wholesale and retail, food manufacturing services and global business services,” he says.
Nazrul says it is crucial to view AI not just as a threat but as an opportunity to ‘ride the wave instead of becoming the victim’.
Nazrul adds that TalentCorp’s study focused on how the 620,000 workers could pivot into new roles, which later led to the creation of the MyMahir platform to address talent challenges and reduce underemployment.
“This platform helps people identify current roles, their transferable skills and the new skills they need to acquire. MyMahir can be the gateway to how we strategise in advancing our nation’s workforce.”
TalentCorp is also working with the Human Resource Development Corporation (HRD Corp) to provide targeted training based on MyMahir findings.
“We want to know how to strategise training programmes and make them accessible through HRD Corp. Several programmes are already ongoing, aligned with industry demand,” Nazrul says.
Socso, meanwhile, runs reskilling programmes under the EIS, with training fees covered up to RM4,000.
“We recommend programmes not only under EIS but also those supported by the Human Resource Development Fund and other institutions,” says Gayathri.
For the future
Meeting the challenges and opportunities posed by AI goes beyond reskilling current workers. The next generation is also being prepared.
Beginning in 2027, the Education Ministry is expected to implement AI technology in school curricula in phases, says its educational planning and research division director Dr Nor Saidatul Rajeah Zamzam Amin.
Saidatul says beginning in 2027, the Education Ministry is expected to implement AI technology in school curricula in phases.
She says the new curriculum will treat technology not just as a tool but as a core competency.
“So we will introduce new subjects related to technology and digital skills from primary to secondary school.”
Saidatul stresses the aim is to promote digital skills for solving real-world problems.
“In developing competencies such as critical thinking, creativity and the ability to work alongside intelligent machines, we hope to prepare students not only to use technology but also to innovate with it.”
The ministry is in the final stage of developing the curriculum.
“To ensure successful implementation, we will reinforce it with comprehensive teacher training. Teachers will be equipped with digital competencies and AI-based pedagogical approaches to integrate AI tools effectively into classroom teaching.”
She acknowledges that the rollout requires significant investment and cross-agency coordination.
“But above all, in terms of education, we believe AI is here to strengthen – instead of disrupt – the system.
Who is most exposed?
ACCORDING to the Novel AI Technologies and the Future of Work in Malaysia study, jobs involving predictable, repetitive or text-based activities with little need for subjective judgment are most exposed to generative AI.
On a scale of 0 to 1 (with 1 being the highest exposure), the top five jobs with an AI automation score of 0.95 include:
> Typing written material from drafts, corrected copies, voice recordings or shorthand using a computer or word processor.
> Checking completed work for correct spelling, grammar, punctuation and formatting.
> Transcribing information from shorthand notes and sound recordings.
> Compiling and processing patients’ medical records, admission and discharge documents and other medical reports for data monitoring and referrals.
> Recording and transmitting buy and sell orders for securities, stocks, bonds or foreign exchange.
JOBS requiring significant human interaction are least exposed to AI takeover, scoring as low as 0.10. The top five low-exposure occupations include:
> Making, ratifying, amending or repealing laws, rules and regulations within a statutory or constitutional framework.
> Providing dental treatment.
> Examining and treating patients, managing pregnancy care and childbirth, inspecting bodies and preparing reports, performing on-call duty and delivering lectures.
> Giving in-flight instruction, supervising solo flights, accompanying students on flight training and demonstrating flight control techniques.
> Massaging clients to improve circulation, stimulate nerves, relieve tension and promote therapeutic effects.