What happens to Southeast Asia’s graduate tech jobs when AI can do the junior work?

For decades, the first few years of a career were supposed to involve doing the work nobody senior wanted to do.

Junior developers fixed relatively simple bugs. Analysts built spreadsheets and prepared reports. Marketing executives wrote first drafts, pulled campaign numbers and put together presentation decks. New consultants researched industries and summarised documents.

The work was not always glamorous, but it served a purpose. People learnt the basics before being trusted with bigger decisions. Artificial intelligence is starting to complicate that model.

Many of the tasks traditionally assigned to graduates are exactly the ones generative AI handles reasonably well. It can summarise reports, analyse datasets, produce first drafts, generate code, prepare meeting notes and research a market in minutes.

---

---

That does not mean Southeast Asia is heading towards the end of graduate employment. In fact, there is little evidence of that happening yet. The bigger question is what happens when businesses still need junior employees but need fewer people to perform the junior work that historically taught them how to become senior employees.

The graduate jobs have not disappeared

The current data presents a more balanced picture than some of the headlines around AI and employment suggest. A July 2026 International Labour Organization study found little evidence that generative AI had caused widespread labour market disruption across ASEAN so far. Employment continued to grow in many occupations considered highly exposed to AI, although the study identified early signs of weaker outcomes for younger workers in some entry-level roles.

The same research estimates that generative AI could affect the working lives of nearly 80 million people across ASEAN. Exposure, however, does not necessarily mean replacement. In many cases, AI is expected to change the tasks within a job rather than remove the job altogether.

Singapore offers a useful example with the number of entry-level professional, managerial, executive and technical job openings actually increasing slightly from 32,500 in December 2025 to 32,800 in March 2026.

Among approximately 18,000 recent graduates from Singapore’s autonomous universities, about 9,100 had found employment by June 2026, while around 3,600 were still looking for work. A further 5,400 were outside the labour force, mainly because they were pursuing further studies or taking a break before starting work.

AI adoption also appears to be changing jobs more frequently than removing them. Among Singapore companies that had adopted AI, only 6% reported reducing headcount and 8% reported lowering hiring activity. In comparison, 19% had redesigned roles and 14% had created AI-related jobs.

Singapore’s Ministry of Manpower also found that 70.7% of companies using AI reported improvements in worker productivity. That distinction matters, because the immediate challenge is less about jobs disappearing and more about the content of those jobs changing.

AI is particularly good at junior tasks

Entry-level workers are potentially more exposed because much of their work involves structured tasks. A graduate analyst may spend hours consolidating information from several reports. A junior marketer might research competitors before preparing a campaign proposal. An early-career developer may be responsible for straightforward coding work.

These are increasingly tasks that can be completed with AI assistance. The World Economic Forum estimates that more than one in three young workers globally are employed in occupations with medium to high exposure to AI-driven changes. It argues that the issue could affect job access, job design, talent pipelines and the way education systems prepare people for work.

For employers, there is an obvious appeal, so that if an experienced employee supported by AI can complete work that previously required several junior team members, the economics of hiring begin to change.

A startup trying to extend its runway may decide it needs three people rather than five. A marketing department may keep the same number of employees but expect considerably more output. A software company may hire fewer generalist developers while competing harder for people with specialist AI, cybersecurity or infrastructure skills.

This fits into a wider restructuring of the technology workforce. As we previously explored, Southeast Asia’s technology employment recovery is becoming increasingly selective, with companies prioritising specific capabilities rather than returning to the broad hiring of earlier technology cycles.

The risk for graduates is therefore not necessarily that there will be no jobs. It is that the bar for getting one could become higher.

The first rung of the career ladder is getting taller

This creates an awkward problem for employers. Companies want graduates who can think critically, communicate with customers, understand commercial priorities and make good decisions.

Those skills usually come from experience. Someone becomes better at analysing a business after producing dozens of analyses. A developer becomes better at reviewing code after writing plenty of it. An account manager learns how to handle difficult clients partly by sitting through uncomfortable client calls.

Junior work has traditionally provided the repetition needed to build that experience. If AI removes too much of the early work, companies may accidentally weaken the process through which employees develop deeper expertise. There is also a danger of expecting graduates to arrive ready to operate at a higher level.

Job descriptions already routinely ask entry-level applicants for internships, industry experience, technical capabilities and familiarity with multiple software platforms. AI could add another expectation: candidates should understand how to work with automation before they even arrive.

Singapore’s own vacancy data gives some indication of the tension around the term “entry-level”. While the country had 32,800 entry-level PMET vacancies in March 2026, nearly 80% required three years or less of work experience.

That may be a fairly broad definition of an entry-level position for someone leaving university with little or no professional experience. AI could push expectations further.

Being good at AI will not be enough

There is an obvious response for graduates: learn to use AI. That is necessary, but it is unlikely to remain a major differentiator for long.

AI literacy may eventually become similar to spreadsheet or search engine literacy. Employers will assume candidates can use these tools rather than treating the ability as a specialist capability.

The more valuable skill will be knowing when the output is wrong. A marketer still needs enough knowledge to recognise weak positioning. A financial analyst needs to question strange assumptions in a model. A developer has to understand whether generated code creates security or performance problems.

A junior employee who can produce ten AI-generated reports but cannot determine which one contains a flawed assumption is not necessarily more useful than someone who produced one manually. This is why AI could make fundamental skills more important rather than less relevant.

Someone who relies on an AI system without understanding the work underneath it can produce more material. That does not necessarily mean they produce better work. Southeast Asia is already confronting this problem at a wider workforce level. As AI moves from pilot programmes into everyday business operations, companies need employees outside dedicated technology teams who can use the tools effectively while understanding their limitations.

The region’s talent problem is therefore changing, because companies still need AI engineers and data specialists. They also need accountants, marketers, salespeople and operations teams capable of applying AI to their existing expertise.

Employers may need to rethink junior jobs

It is easy to look at AI as a way to remove repetitive work and reduce costs. It is harder to consider what happens to talent development five years later.

If businesses automate most entry-level analysis, research and execution, they still need a way to train the managers, specialists and leaders they will require in the future. That could change the structure of graduate programmes.

Instead of spending the first year preparing routine reports, junior employees could move into client meetings earlier. They might spend more time checking AI output, interpreting information, running experiments or working across several parts of the business.

Managers may also need to become more deliberate about teaching. Historically, junior employees could learn through repetition. Complete the same type of task enough times and some understanding develops naturally.

AI removes some of that repetition.

Managers may therefore need to explain why decisions are being made rather than simply giving employees enough work to eventually work it out themselves.

This could create better entry-level jobs. Removing administrative and repetitive tasks should give graduates more time for work requiring judgement, creativity and communication

Southeast Asia cannot afford to get the transition wrong

The issue matters particularly for Southeast Asia because the region has a large young workforce and economies trying to move more workers into higher-value digital jobs.

At the same time, companies throughout the region are being encouraged to automate. Both objectives can coexist, but the transition requires more thought than simply teaching graduates how to write prompts.

There is already strong demand for new approaches to workforce development. Learning platforms are increasingly being judged on whether they can connect training directly with employability and workplace outcomes.

Universities face a similar problem. They need to reconsider what students should learn when generating a first draft, basic analysis or a straightforward piece of code becomes almost effortless.

Employers need to decide which entry-level tasks remain important for learning even when AI could complete them faster. Graduates need to develop judgement and domain knowledge alongside their technical skills. AI may eventually remove some junior positions. The evidence across ASEAN suggests that this is not happening at scale yet. The ILO’s latest assessment still finds limited evidence of widespread labour market disruption.

What already appears to be changing is the definition of being junior. The next generation of graduates may enter companies with fewer repetitive tasks, better tools and much higher expectations from their first day.

That could produce a more productive Southeast Asian workforce. It could also leave young workers without enough space to learn how the work is actually done.

The difference will depend on whether companies treat entry-level employees primarily as a cost to automate or as the experienced employees they will still need to develop.