A reflection on why human-centred AI in Thailand must move beyond individual preference to recognise the obligations, relationships and connected responsibilities that give personal choices their meaning.
Personalisation begins with a simple question: what does this person prefer? Which product do they usually choose? What content keeps their attention? Which route, schedule or recommendation best matches their past behaviour?
These questions are useful because preference can often be observed through data. But Thai decisions are not always organised around personal preference alone. A person may choose a job because it allows them to remain near ageing parents. A consumer may spend on the family before buying something for themselves. A senior employee may stay in a difficult role because colleagues depend on them. A business owner may continue supporting a long-term partner even when another supplier offers a lower price.
From a purely individual model, some of these choices may appear irrational. From a relational perspective, they express obligation, gratitude, identity or responsibility.
Artificial intelligence may struggle with this difference. A system sees that a person repeatedly rejects the highest-paying opportunity — it may not understand the family responsibility behind the choice. It sees a customer buying for several different age groups and interprets inconsistent preferences — it may not recognise that one account is serving an entire household. It sees an employee declining relocation and predicts low ambition, when the decision reflects a responsibility the organisation has never asked about.
My doctoral research considered cultural dimensions including individualism, long-term orientation, relationships, loyalty and obligations. Its broader contribution was to show that technology cannot be assumed to operate outside cultural context.
The intelligent age may intensify individualisation because personal data is usually attached to one user. One account. One profile. One predicted preference. But human life is often lived through connected responsibilities.
This has important implications for Thai services and organisations. A healthcare system designed around the individual patient may still need to recognise the family members who participate in care. A financial assistant may need to understand household obligations rather than optimising only personal consumption. A workplace AI assessing mobility, availability or career ambition should not reduce a culturally embedded decision to lack of motivation.
This does not mean every obligation is healthy. Family expectations can become excessive. Loyalty can prevent necessary change. Gratitude can be used to demand compliance. People should retain the ability to choose a life beyond inherited responsibility. But autonomy does not require pretending that relationships have no legitimate claim upon us.
Human-centred AI in Thailand should therefore be capable of asking: Who else is affected by this decision? Which commitments is the person trying to preserve? Is the recommendation optimising for an isolated user who does not actually exist?
The future of personalisation should not be limited to knowing what we like. It should leave room for understanding what we care about enough to carry — even when carrying it is inconvenient.
A system may understand what we prefer, yet still misunderstand us completely if it cannot see the people and obligations that give our choices meaning.