In short
- AI is more likely to change public sector jobs than remove them, but the outcome depends on choices governments make now.
- An ERIA researcher argues AI could create or destroy public sector jobs in ASEAN, depending on whether it supports workers or replaces them.
- Routine, easily automated tasks are the most exposed. Lower-skilled staff face the greatest risk if they cannot move into more complex work.
- Be careful with big numbers. The often-quoted 6.6 million jobs across the ASEAN-6 by 2028 comes from a 2018 study of technology adoption in general, not AI specifically.
- Five steps help: map tasks, redesign roles, reskill before you automate, build career paths, and involve staff early.
Every government in ASEAN is asking the same question: if we bring in AI, what happens to our people? The honest answer is that nobody knows yet, and the answer will depend on what governments choose to do. This article looks at what the evidence says, what is uncertain, and five steps that give staff the best chance of moving up, not out. It is written for ministry and agency leaders, heads of HR, and anyone planning a workforce through AI adoption.
The short answer
Writing for ERIA, a researcher argues that AI could either create or destroy public sector jobs in ASEAN. Which one happens depends on whether governments develop and integrate AI so that it supports human workers rather than replacing them. Because AI adoption is widely seen as unavoidable, the recommended response is to regulate and integrate it, and to reskill and upskill public servants for higher-skilled work that is harder to automate. That includes broader job roles and clearer career paths.
What the evidence says, and what it does not
- Adoption is early but widespread. ERIA cites a Kearney study finding early-stage AI adoption in 83% of Southeast Asian countries, with over 70% of survey respondents viewing AI as vital for the future.
- Governments are already moving. ERIA points to Singapore’s National AI Strategy (2019) and its One Service chatbot for municipal complaints, an AI-based online public service platform in Indonesia, and Viet Nam’s National AI Strategy for 2021 to 2030.
- Skills change fast. In January 2026, ACCA’s Asia Pacific director said job roles are evolving within 2.5 to three years, and listed data analytics, cybersecurity, digital literacy and ESG reporting among the most at-risk skills by 2030.
- The big job-loss number needs care. ERIA cites a 2018 Cisco and Oxford study projecting that increased technological adoption will affect around 6.6 million jobs across the ASEAN-6 economies by 2028. That figure covers technology adoption broadly, not AI specifically, it is several years old, and the ERIA piece gives no comparable figure for jobs created.
So the honest summary is that the direction is clear, the scale is uncertain, and the outcome is shaped by decisions still to be made.
Which work is most exposed
ERIA points to the automation of routine, easily automated tasks. As illustrations, not forecasts, think of data entry, form processing, routine correspondence, scheduling and first-line enquiries. Roles built mainly on such tasks are the most exposed. ERIA also warns that lower-skilled public servants face a higher risk of unemployment if they cannot move into more complex roles.
Where new work may come from
The same source argues that automating repetitive work can free public servants for more complex tasks, which can build analytical and innovation skills and create demand for IT specialists, engineers and data scientists.
Five steps that tip the odds in your staff’s favour
1. Map tasks before you buy
For each role, list the tasks and how much time each takes. Then sort each task into one of three groups: automate (routine and rules-based), augment (a person leads, AI helps) or elevate (human work that grows in value).
| Task (illustrative) | Automate, augment or elevate | What changes |
|---|---|---|
| Logging routine enquiries | Automate | The tool handles the log; a person handles exceptions. |
| Drafting a first reply | Augment | A person edits and approves. |
| Handling a complex case | Elevate | More time to do this well. |
2. Redesign roles, not only tools
When tasks change, jobs should change with them. Merge narrow roles into broader, cross-functional ones, and update job descriptions so people know what they are now responsible for.
3. Reskill before you automate
Start training before a tool arrives, not after. Begin with AI literacy for everyone, then build skills by role. Our guide to building an AI-ready civil service sets out the sequence.
4. Build career paths
People will take reskilling seriously when it leads somewhere. Create clear routes from today’s roles to the new ones, and consider linking micro-credentials to progression or pay.
5. Involve staff early, and be honest
Set up a steering group that includes the people affected. Explain what will change, what will not, and how decisions will be made. Where you can truthfully give an assurance, such as how any role reductions would be handled, give it early. Silence breeds fear.
What to tell your staff
- What is changing: the tasks that will be automated or supported, and when.
- What is not changing: what people will still own, and how their judgement matters.
- What support exists: the training, the time to learn, and the routes to new roles.
Warning signs that a plan is going wrong
- A tool is rolled out with no change to roles or workloads.
- Training reaches managers but not the people whose tasks change.
- Staff first hear about the change from the tool, not from leaders.
- There is no measure of whether people moved into better roles.
Frequently asked questions
Will AI take civil service jobs?
It will change many of them. Whether it removes jobs depends on how governments introduce it. If AI supports workers and staff are reskilled for more complex work, roles shift more than they disappear. If it is used mainly to cut headcount without reskilling, job losses are more likely.
Which public sector jobs are most at risk?
Those built mainly on routine, easily automated tasks, and those held by staff who lack a route into more complex work. Mapping tasks, not job titles, is the way to see where exposure really sits. Start by listing what each person does in a typical week, then sort each task.
How should governments reskill staff?
Start before the tool arrives. Give everyone AI literacy, add skills by role, build a small group of champions, and link new skills to career paths. Measure what people do at work, not only who attended. Tell staff early what support exists, so reskilling is seen as a route forward.
What to do next
Pick one team and map its tasks this month. You will see quickly where AI could help, where roles need to change and what skills to build. If you would like help with job redesign, reskilling or change planning, discuss a project with Blak Rice. See our government work and project consulting.
Sources
- ERIA, “AI on ASEAN’s public sector jobs: a helping hand or a threat to employment?” (30 January 2024, an opinion piece by an ERIA research associate; includes the cited Kearney and Cisco and Oxford figures): eria.org
- Bernama, “ASEAN Must Accelerate TVET Transformation As Digital, Sustainability, Skills Gaps Widen - ACCA” (13 January 2026): bernama.com