Pacific Money

Mind the Gap: How Southeast Asia Can Make the AI Leap

Recent Features

Pacific Money | Economy | Southeast Asia

Mind the Gap: How Southeast Asia Can Make the AI Leap

Artificial Intelligence technology holds out huge potential benefits for ASEAN – if the region can close some crucial shortfalls.

Mind the Gap: How Southeast Asia Can Make the AI Leap
Credit: Pixabay/Gerd Altmann

Southeast Asia is rapidly experiencing an artificial intelligence (AI) awakening. A study by EDBI and Kearney on the state of AI readiness in Singapore, Malaysia, Thailand, Vietnam, and the Philippines revealed an increasing momentum on the adoption of AI use-cases in various industries. Indonesia’s recent launch of a National Strategy for AI exemplifies this growing realization about the potential economic benefits of AI for the region. If Southeast Asia gets AI right, it could add $1 trillion to its GDP by 2030.

But underneath the spectacle and promise of AI in Southeast Asia, embracing this new technology requires a deeper degree of introspection. The region must first confront the three debilitating challenges it continues to ignore, which are integral to its emergence and advancement as a data-driven economy: inadequate investments in research and development (R&D); fragmented data governance; and weak cybersecurity.

First, a cursory glance at the R&D expenditure on AI in Southeast Asia shows a stark divide. On this front, Singapore leads the pack. Aiming to transform itself into the region’s AI hub, the city-state’s R&D investments totaled $68 per capita in 2019. In contrast, the Philippines, Vietnam, Malaysia, and Indonesia had committed less than $1. 

In the current value chain of machine learning, one of the subfields of AI, Southeast Asia is falling behind. The U.S., Japan, China, the U.K., Germany, and South Korea occupy the top-tier spots in terms of algorithmic training able to harvest data in order to make sophisticated predictions about the future. Meanwhile, Singapore, Indonesia, and the Philippines are relegated at the lower end of the value chain, characterized by capital- and labor-intensive tasks of data collection, data storage, and data preparation. The U.S. is unmatched in algorithmic training due to its computing power, enabled by bespoke semi-conductors, followed by Japan, South Korea, Germany, and the United Kingdom.

For Southeast Asia to elevate its competitiveness in the algorithmic training node of the AI economy, boosting investments in AI research, computing technology, and human capital development within the context of public-private partnerships must become national priorities. 

With the exception of Singapore, which is already doing this, the Philippines, Vietnam, Malaysia, and Indonesia must all prioritize investments in AI research and computing infrastructure. National governments must coordinate with the private sector, which is overwhelmingly leading AI R&D, to ease any regulatory bottlenecks and identify incentive mechanisms for innovation.

On human capital, higher education departments and ministries should develop working partnerships with native digital companies. Establishing AI laboratories or centers in various universities, supported by tech companies, is a viable route toward producing highly qualified AI talents. Supporting pilot projects among the start-up community could also spark local innovation and collaboration.

The promotion of Science, Technology, Engineering, and Mathematics (STEM) education should emphasize not only the technical and analytical know-how, but also the soft skills necessary to consider the human and ethical dimensions of AI technology. Businesses and organizations in Southeast Asia will benefit from the reservoir of talents who are capable of indigenizing algorithmic training and AI models that reflect the local context. They will be far more effective in eliminating systemic bias and producing credible results for various AI use-cases and applications.

Second, Southeast Asia must deal with its fragmented approach to data governance if it is to build a robust AI ecosystem. Due to rapidly growing internet penetration through social media apps and e-commerce platforms, Southeast Asian nations have been generating large amounts of data. Recognizing this reality, ASEAN in 2018 unveiled its Framework on Digital Data Governance, which aims to “enhance data management, facilitate harmonization of data regulations among ASEAN Member states and promote intra-ASEAN flows of data.”

However, the brewing divide over the cross-border flow of data versus data localization might affect ASEAN’s progress in achieving an interoperable data governance blueprint. Differing views on national security and the economic benefits of data management are becoming contentious issues among ASEAN member states.

The patchwork of data governance preferences in Southeast Asia will impact the harmonization, extraction, and flow of data, which consequently create additional costs for data storage, data processing, data retrieval, and analysis. If this persists, it will constraint AI collaboration and integration at the national and regional levels, both of which are central to building an AI ecosystem.

To move forward, Southeast Asia must adopt a risk-based approach toward cross-border data flow using appropriate safeguards that are commensurate to the possible risks involved in transferring data. By institutionalizing mutual recognition agreements, Southeast Asia can identify specific conditions that determine the permissibility of data transfer rather than just adhering to blanket policies on data protectionism.

Third, there is the question of cybersecurity. Southeast Asian nations are prime targets of cyberattacks. With cybersecurity spending lagging, the region could lose an estimated $180 billion to $365 billion in the next five years from massive data breaches.

Malicious actors can attack AI systems through data poisoning, tempering of categorization models, and exploiting backdoors. These methods can compromise the learning ability and gain control of AI systems. The introduction of erroneous datasets or the exploitation of backdoors in neural networks can manipulate inputs that could cripple the system’s reliability. Moreover, the advent of AI-infused cyberattacks makes it even harder to detect such anomalous activities.

As Southeast Asia begins to embrace AI, bolstering cybersecurity standards for government agencies and departments, technology companies, and universities is highly critical. Adopting accountability mechanisms and liability regimes is also vital in order to ensure that all parties involved in the design and development of AI observe stringent auditing and testing standards. In countering adversarial AI attacks, businesses and organizations must upgrade to investing in AI-infused cybersecurity. 

Despite its uneven digital maturity, Southeast Asia is well primed to become an emerging data-driven economic powerhouse. The current dynamism surrounding the advent of AI exemplifies the growing interest in the region to reap the rewards of the fourth industrial revolution. But rather than just riding on the hype of AI, Southeast Asia must first have a clear-eyed assessment of its comparative strengths and inherent weaknesses. This necessitates bridging the existing gaps in R&D, data governance, and cybersecurity. All this will be necessary before the region can undertake a full AI leap.

Mark Manantan is currently the Lloyd and Lilian Vasey Fellow at the Pacific Forum and a non-resident fellow at the Center for Southeast Asian Studies in Taiwan. He is the founder and strategic director of Bryman media.