Localizing AI through data, legal frameworks, and training programs is pivotal for Southeast Asia to not only access but also master AI technology. Simultaneously, this strategy helps preserve and promote cultural values in the digital age.

AI: Opportunity for Breakthrough or Catalyst for Deepening Inequality?

While global technology corporations are accelerating the development of large language models (LLMs), promising significant leaps in productivity and efficiency, the digital divide between countries, and even between businesses within the same region, is widening.

This is the warning highlighted by Professor Soumitra Dutta – a member of the VinFuture Prize Council and Dean of the Saïd Business School at the University of Oxford – at the online seminar “AI Without Borders: Ethics and Technology” co-organized by the UNESCO Office in Vietnam and the VinFuture Foundation. He called AI an “amplifier of existing inequalities.”

According to Professor Soumitra Dutta, AI will further widen the gap between businesses with strong digital foundations and those limited in infrastructure, human resources, and data. This risk is particularly concerning in Southeast Asia, where the disparity in technological capabilities between large enterprises and small and medium-sized enterprises (SMEs) remains significant. Professor Soumitra Dutta proposed three strategic “buckets” for every nation and business: creation, regulation, and adoption. Only by synchronously building all three stages – from developing indigenous algorithms and establishing appropriate legal frameworks to practicing organizational transformation – can AI avoid becoming a “profit-generating tool” for those already at an advantage.

For businesses considering adopting AI, the message is clear: it is not enough to simply “buy technology” from external sources. Instead, it is necessary to proactively invest in internal capabilities, from data and human resources to processes, to ensure that AI serves the right business objectives and does not make the business dependent on global platforms developed for very different markets. When implementing a suitable AI strategy and localization, AI will both support operations and act as a lever to help businesses rise in an increasingly competitive and volatile market.

Crafting AI Policy from Local Realities

If technology can create inequality, then the policy framework is the tool to regulate and narrow that gap. However, in Southeast Asia – a region rich in diversity but lacking a sufficiently strong voice in the global AI race – the challenge of localizing AI lies not in copying advanced laws, but in building regulations flexible enough to reflect unique cultural characteristics, institutions, and development needs.

Sharing insights from the Global South, Mr. Jose A. Guridi – a UNESCO consultant on artificial intelligence (AI) directly involved in assessing AI readiness in many developing countries – emphasized that directly copying the AI Act or GDPR could stifle innovation if feasibility is not considered.

Instead, it is necessary to build a flexible legal framework focused on priorities such as smart agriculture, telemedicine, and inclusive finance. Mr. Jose proposed a localized policy approach, focusing on addressing practical issues such as enhancing productivity in smart agriculture, improving access to telemedicine, and expanding inclusive finance for vulnerable communities.

Such policy orientations, according to him, will not only help Southeast Asian countries move in the right direction for development but also ensure that AI truly serves people, especially vulnerable groups in society. For the business community, Mr. Jose’s advice is very clear: instead of passively observing, businesses should actively participate in the early stages of AI policy development and experimentation.

This will not only help ensure that regulations are aligned with business realities but also open opportunities to shape new standards where domestic companies can become leaders in developing responsible and culturally appropriate AI applications. In other words, the “law-making” process is no longer just a matter for governments or legal experts but a strategic playing field where businesses can and should have a voice if they want AI to become a sustainable part of their long-term growth strategy.

Decoding Cultural Barriers

Returning to the local context, Mr. Adya Danaditya – a UNESCO consultant on artificial intelligence (AI) and Product Group Lead at INA Digital Edu, which develops digital solutions for the Indonesian Ministry of Education – painted a detailed picture of how culture and language can hinder any AI application efforts in Southeast Asia.

First and foremost is the language barrier. “Most large language models like ChatGPT are trained primarily on English data,” Mr. Adya Danaditya shared. When businesses ask AI to generate code or draft content, the English output is often complete and detailed; while the Indonesian version only summarizes the main points, and the Lao version even completely fails to execute the request.

The result of this imbalance is not only higher integration costs but also a degraded customer experience and the need for constant manual intervention by operations teams. The next issue is “epistemic injustice.” When local users search for information about local festivals, cuisine, or customs, AI often only returns superficial or incorrect information, even confusing unique cultural values, while Western topics are analyzed in detail. Consequently, the younger generation may form a distorted understanding of their own heritage, threatening the sustainability of long-standing cultural values.

Regarding economic impact, Mr. Adya Danaditya pointed out that the creative economy in Indonesia currently attracts nearly 16 million workers, from graphic design to digital content production. The widespread use of AI-generated art, along with creative automation algorithms, is “depleting” opportunities for local workers, especially women and informal workers in smaller economies like Laos, as policies to protect employment have not yet been completed. To address these linguistic and cultural gaps, Mr. Adya Danaditya likened the preparation of data, the preservation of heritage, and user training to a cultural “vaccine” protocol. First, building a high-quality local data repository is a crucial step: through initiatives like SEACROT, businesses and research organizations can establish multilingual datasets, helping AI models not only understand the correct meaning but also accurately reproduce local cultural characteristics.

Next, digitizing and publicizing cultural archives, from folk tales and traditional festivals to unique customs, will create a reliable source of original data while fostering a sense of heritage preservation within the community. Finally, Mr. Adya Danaditya emphasized the importance of localized AI literacy programs. Instead of just deploying new tools, businesses need to organize training sessions and provide AI usage instructions in the native language for all departments, from warehouse staff and customer service teams to product development groups, to enhance interaction capabilities, oversight, and the development of AI applications suitable for local realities.

When these three elements are seamlessly combined, AI can truly become a catalyst for promoting creativity and sustainable growth for Southeast Asian businesses, rather than an invisible barrier that further entrenches historical injustices. Only when these “vaccine doses” are fully administered, from standardized data content and supportive regulations to practical training programs, can AI truly transform into a tool that serves businesses and communities in Southeast Asia, instead of becoming an invisible barrier that deepens existing inequalities.