This article is part of a series revisiting ideas from my doctoral research, Characteristics of Culture in the Context of Thai Society and Virtual Communities, through the lens of today's AI-enabled society.
Artificial intelligence promises a deeply personalised future. It can adapt to our language, preferences, working style and interests. It can remember context, recommend content and present information in ways designed specifically for each individual. At first, personalisation appears to be the opposite of standardisation.
But there is a paradox. The experience may be personalised while the underlying intelligence remains culturally concentrated. Many of the world's most influential AI systems are developed using datasets, assumptions and design priorities that do not represent every language, society or worldview equally. Research and policy discussions continue to highlight the disproportionate influence of English-language and high-income-country content in modern AI.
UNESCO has consequently called for greater geographical, cultural and linguistic diversity in AI training data, together with stronger transparency concerning the cultural effects of algorithmic systems.
This returns us to the question behind my original research. Will technology create a single global culture, or will local cultures adapt technology while retaining their identities? With AI, the answer may depend on design choices being made today.
If intelligent systems become the main channel through which people learn, write, search and create, the values embedded in those systems can quietly influence language and thought. Cultural expressions that are richly represented in the data may be amplified. Those that are poorly represented may become simplified, misinterpreted or gradually less visible.
The danger is not that AI will deliberately destroy local culture. The greater risk is that it will make cultural uniformity convenient.
When one communication style is consistently presented as professional, people may stop recognising alternative forms of professionalism. When one narrative structure is treated as the ideal, different traditions of storytelling may be regarded as less sophisticated. When one approach to leadership is repeatedly recommended, locally grounded models may appear outdated even when they remain effective.
For Thailand, the opportunity is not to reject global AI systems. It is to participate more actively in shaping them. Thai organisations, universities and technology developers should invest in high-quality Thai-language datasets, local knowledge structures and culturally informed evaluation. AI should understand more than the literal Thai language. It should recognise context, indirect communication, social relationships, seniority, consideration for others and the importance of preserving dignity or "face."
At the same time, cultural sensitivity should not become cultural confinement. Not every Thai individual behaves according to a national average. My research emphasised the importance of distinguishing culture at the societal level from personality at the individual level. A national cultural characteristic must never become a stereotype applied to every member of that society.
The goal should therefore be culturally aware personalisation, not cultural profiling. A culturally aware system understands that context matters. A culturally profiling system assumes that nationality determines the individual. This distinction will become essential for responsible AI.
The best future is neither one universal intelligence that makes everyone think alike nor isolated national systems that prevent the exchange of ideas. It is a connected intelligence capable of respecting difference while enabling collaboration. Technology should help cultures communicate — not require them to disappear.
True global intelligence is not intelligence that speaks with one voice. It is intelligence capable of listening to many.