Is Southeast Asia preparing students to build with AI or just prompt chatbots? Credit: Unsplash/Cash Macanaya
ASEAN’s AI Challenge Lies in Building Practical Capability
16 November 2026/6 Minutes of Reading
Introduction
Southeast Asia is expanding its use of AI, but familiarity with it does not necessarily translate into the ability to develop useful applications. The challenge for ASEAN is turning AI education into practical skills while addressing inequalities across its education systems.
Across Southeast Asia, the debate over AI in education is often framed around whether students should be encouraged to use tools such as ChatGPT. However, this question overlooks a more consequential policy problem: does using AI actually prepare students to participate in the region’s developing AI economy?
The distinction matters because ASEAN’s ambitions extend beyond adopting new technologies.
The ASEAN Digital Masterplan 2030 identifies a growing shortage of skilled professionals in AI, cybersecurity, cloud computing and data science. At the same time, education systems across the region operate with different levels of digital infrastructure, teacher preparedness and technical capacity.
These conditions create a gap between AI adoption and AI capability. Students may become familiar with generative AI without learning how to assess its reliability, integrate it into applications or understand the data and systems behind its outputs.
Closing this gap requires more than introducing AI tools into classrooms; it requires examining what students are actually learning, whether they can apply those skills and how effectively education policies prepare them for technological change.
Exposure ≠ Capability
The first challenge is bridging the gap between AI exposure and practical capability. The scale of AI’s potential impact on employment makes this distinction increasingly important.
Around 22.9% of ASEAN employees (equivalent to nearly 80 million workers) are in occupations with more than minimal exposure to generative AI, although only 3.3% of workers are in the highest exposure category.
This is not to argue that the jobs of these 80 million workers will disappear. Occupational exposure measures the potential for tasks to be affected by AI, not the certainty of job displacement.
Indeed, the International Labour Organization (ILO) – which conducted the study – found little evidence of widespread labour-market disruption so far. Its findings, instead, point to the importance of preparedness, skills development and policy decisions in determining how AI affects employment.
For education policymakers, the distinction between exposure and capability is equally important.
Being able to generate an essay, summarise a document or obtain an explanation from an AI assistant demonstrates familiarity with a tool. However, these do not necessarily demonstrate an understanding of how to verify its outputs, recognise systematic errors or apply the technology to a problem that requires independent judgement.
Developing an AI application involves a different set of competencies. Beyond using an existing model, developers must define the problem, select suitable inputs, assess output quality, handle data responsibly and test whether the system performs reliably. An application can produce convincing responses while still giving inaccurate feedback or failing to meet its intended purpose.
My experience in developing MyBertutur, an AI-powered speaking practice platform for Malaysian students, has made this distinction particularly relevant to education.
Integrating AI into a learning product is only one part of the process. The feedback must also be useful for the learner, understandable in context and sufficiently reliable for its intended purpose. A generated response is not automatically a meaningful learning outcome simply because it sounds convincing.
The broader implication is that AI adoption figures alone cannot establish whether a country is developing AI capability. The number of students using AI tools may indicate growing access, but it does not show whether they can evaluate the technology or create useful applications with it.
Therefore, for ASEAN, the policy challenge is to distinguish between digital consumption and the development of skills that allow people to apply technology critically and productively.
Inequality
The second challenge is that students across Southeast Asia do not begin with the same opportunities to develop these skills.
A study found that students from the region’s wealthiest households are almost eight times more likely to have internet connectivity at home than students from the poorest households. Meanwhile, only around half of rural primary schools in ASEAN are connected to the Internet.
Although this relates to broader digital education challenges rather than AI adoption alone, it highlights an underlying inequality that remains relevant to AI policy. Access to an AI application is of limited value when students lack reliable connectivity, suitable devices or teachers who can guide them in using it critically.
Evidence on AI preparedness reinforces the concern. The ASEAN Foundation reported that 72.4% of students and 70.5% of educators across ASEAN had never taken an AI course. The report also found that many respondents use AI mainly for basic tasks such as searching and writing, with differences between urban and rural settings.
These findings suggest that AI education cannot be treated as a uniform intervention. A school introducing AI to students for the first time faces a different challenge from one that already offers computing lessons, coding activities and opportunities to develop software. Applying the same curriculum without accounting for these differences risks producing uneven outcomes.
Singapore illustrates what a more structured approach can look like. Infocomm Media Development Authority’s AI for Fun programme for secondary schools includes hands-on activities involving large language models, custom chatbots, training-data quality and bias in AI-generated content. Students also explore how generative AI can be applied to practical problems.
The significance of this approach is not simply that students are introduced to AI. It connects use with an understanding of how the technology works, where it can fail and how it might be applied. However, reproducing this approach elsewhere would require attention to local conditions, including teacher training, curriculum priorities and access to digital infrastructure.
For ASEAN, this makes regional cooperation more complicated than distributing a common set of AI learning materials. Shared educational objectives are useful, but their implementation must account for differences in national capacity. Without that distinction, regional AI initiatives could expand access while leaving existing disparities in practical skills largely intact.
Measuring Progress
The third challenge concerns how AI education programmes are evaluated.
Enrolment, course completion and the number of people reached are useful indicators of programme scale. However, they do not, by themselves, establish whether participants have developed the ability to apply AI effectively.
This distinction is relevant to ASEAN’s existing initiatives. The AI Ready ASEAN programme, supported by Google, aims to expand AI literacy across the region through educator training, community learning and accessible educational resources. Such programmes can help establish a foundation of AI understanding, particularly among groups that have had limited exposure to the technology.
The next question is what participants can do after completing the training. Can students identify when an AI-generated answer is unreliable? Can they explain why a model may produce biased results? Can they develop and test a simple application that addresses a real problem? Can teachers design assignments that assess independent reasoning rather than merely the quality of an AI-generated response?
These questions provide a more informative basis for evaluating educational outcomes. Programmes could assess students’ abilities before and after training, use practical projects to test what they have learnt, and measure whether teachers can incorporate AI into subject-specific learning. Results should also be examined across different school environments to determine whether the programmes are reaching students with fewer resources.
Malaysia’s recent policy direction reflects the distinction between using AI and applying it to practical problems.
In an August 2026 speech at AI Malaysia Takeover, Malaysia’s Minister of Digital, Gobind Singh Deo, emphasised the importance of developing people who can move beyond consuming AI technologies towards creating applications and solutions. The government has also outlined an AI Builders Hub under Rakyat Digital to provide opportunities for Malaysians to experiment with AI and develop practical solutions.
The initiative illustrates the importance of connecting learning with application. Its longer-term value, however, will depend on whether participants develop demonstrable skills and can use them beyond the training environment.
At the regional level, ASEAN could build on existing education and digital cooperation by encouraging member states to share practical assessment methods, teaching resources and examples of student-led AI projects. A common framework for measuring foundational AI literacy and applied capability would also make it easier to compare progress without assuming that all countries have identical starting conditions.
Such an approach would shift the focus from how many people have attended AI training to whether that training has improved their ability to understand, evaluate and apply the technology. It would also help policymakers identify which interventions work in different contexts, rather than treating participation figures as evidence of success.
Conclusion
The central question for ASEAN is not simply whether students should be encouraged to use AI; it is what kind of capability that use is expected to develop.
The evidence points to three connected concerns: 1) exposure to AI does not automatically produce practical skills; 2) unequal educational conditions affect students’ ability to acquire those skills and; 3) programme participation alone cannot demonstrate whether learning translates into capability.
These concerns require a more precise approach to AI education. Access to tools and basic literacy remain important, but they should be accompanied by opportunities to evaluate outputs, understand limitations and apply AI to real problems.
Regional cooperation can support this process, provided that shared objectives are adapted to the different circumstances of ASEAN member states.
Ultimately, the distinction between AI adoption and capability is a distinction between using technology and being able to make informed, effective use of it. For ASEAN, closing that gap will be an important test of whether its AI ambitions translate into a more capable workforce and a stronger regional digital economy.
The views expressed are those of the authors and do not necessarily reflect those of STRAT.O.SPHERE CONSULTING PTE LTD. This article is published under a Creative Commons Licence.
Republications minimally require: 1) credit authors and their institutions, and; 2) credit to STRAT.O.SPHERE CONSULTING PTE LTD and include a link back to either our home page or the article’s URL.


