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.
Human identity has always been shaped through relationships. We learn who we are partly from how family, colleagues and society respond to us. We adopt roles, values and behaviours through repeated interaction with the communities around us.
Digital environments introduced a new dimension to this process. People could construct profiles, select what to reveal and participate in multiple communities under different identities. The same individual could behave differently in professional, political, entertainment or anonymous online spaces.
Artificial intelligence intensifies this process because it does not simply observe identity. It predicts it.
Algorithms classify people according to their interests, behaviours, networks, purchasing patterns and emotional responses. They decide which content is relevant, which products are suitable and which version of the digital world each person is most likely to engage with.
Over time, these predictions can begin to influence the person being predicted. When a system repeatedly shows someone the same kind of content, it strengthens certain interests while leaving others unexplored. When it identifies a person as politically conservative, highly anxious, price-sensitive or interested in a particular lifestyle, it may continue presenting information consistent with that classification.
The recommendation becomes reinforcement. The profile begins to shape the person.
This raises a deeper question: are intelligent systems discovering who we are, or gradually narrowing who we are allowed to become?
Personalisation is often presented as convenience. It reduces irrelevant information and helps people find what they are likely to value. But identity is not static. Human beings are capable of contradiction, experimentation and change. We develop by encountering ideas that do not fit our previous behaviour. We discover new interests by being exposed to something unexpected. We revise our beliefs because another person challenges the story we tell about ourselves.
An algorithm trained primarily on past behaviour may become very effective at predicting yesterday's identity. It may be less effective at supporting tomorrow's growth.
This is particularly important for younger generations whose sense of self is still developing. When recommendation systems continuously reflect a narrow version of their interests, fears or insecurities, those systems may influence identity before the person has had the opportunity to explore it independently.
The danger is not simply that algorithms misunderstand us. The greater danger is that we may begin to understand ourselves through the categories algorithms assign to us.
In business, similar patterns occur when AI systems categorise employees according to performance data, communication style or predicted potential. These classifications may improve decision-making, but they can also create self-fulfilling expectations. A person labelled as high-potential may receive more opportunity. Someone classified as unlikely to succeed may receive less support and therefore become less likely to succeed.
Intelligent systems should help people understand patterns in their behaviour without turning those patterns into permanent identities. Good personalisation should leave room for surprise. Good leadership should recognise that data describes behaviour in a particular context, not the total potential of a human being.
The intelligent age will require us to preserve the right to become someone different from the person our data predicts.
A human being is more than a pattern of past behaviour. Intelligence should help us grow beyond our data, not imprison us within it.