This reflection extends Thailand's Human Future — the responsibility of public institutions to learn before they scale automated decisions across citizens' lives.
Government systems carry enormous responsibility. They manage taxation, licensing, healthcare, education, welfare, justice and access to essential public services. When these systems operate slowly, citizens experience frustration. When they operate unfairly, citizens experience something more serious: loss of trust.
Artificial intelligence offers an opportunity to improve public administration. It can help process applications, identify service needs, detect unusual patterns and make government information easier to access.
But automating an outdated institution does not automatically create a modern institution. It may simply allow the same weaknesses to operate faster. A confusing rule can become an automated confusing rule. A biased process can become a scalable biased process. A department that does not listen to citizens may use AI to answer more questions without understanding why people remain dissatisfied.
Before public institutions automate, they must learn. They must understand how citizens actually experience the service — not only how the process appears in an official procedure manual.
This connects with one of the practical implications of my research: leaders of technology-mediated communities should consult members, provide understandable rules and make it possible for junior participants to contradict authority without fear. Applied nationally, this means public-sector AI should be developed with citizens, frontline officers and affected communities — not only technical vendors and senior administrators.
People closest to the problem often see what management cannot. A local officer may understand why a digital form repeatedly fails. A citizen may identify a consequence invisible in the policy model. A smaller community may reveal that the national data does not represent its circumstances.
Thailand also needs public institutions capable of admitting when an automated system is wrong. There must be clear responsibility, meaningful appeal and authority to correct outcomes. Technology should not become another layer through which nobody feels able to help.
A government should not automate a process before it understands the people who must live with its consequences.