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 societies depend on trust. We trust institutions, experts, colleagues and systems because it is impossible to verify everything ourselves. Modern life would become unmanageable if each person had to personally inspect every medical diagnosis, financial transaction, engineering calculation or legal interpretation.
Artificial intelligence creates a new kind of trust relationship. We are increasingly asked to accept recommendations produced by systems whose reasoning may be difficult to understand — even for the people who built them.
An AI system may recommend who should receive credit, which employee should be promoted, which patient requires urgent attention or which business strategy has the greatest probability of success. The recommendation may be statistically strong. But statistical performance does not automatically create legitimate trust.
There is a difference between trusting a system because we understand its strengths and limitations, and trusting it because it usually sounds confident.
The fluency of generative AI makes this distinction especially difficult. Human beings often associate clear language with clear thinking. When an AI gives a structured explanation in a calm and authoritative tone, it may appear more reliable than it actually is. Confidence becomes a substitute for evidence.
In traditional hierarchical environments, people may already hesitate to question authority. When technological authority is added, the hesitation can become stronger. A manager may say: “The system recommends this.” The statement can end discussion even when no one fully understands how the recommendation was produced.
This creates what might be called algorithmic distance. The decision affects real people, but responsibility appears to belong somewhere inside the system. The employee blames the manager. The manager blames the algorithm. The organisation blames the data. Everyone participated, but no one feels fully responsible.
Trustworthy AI therefore requires more than accuracy. It requires appropriate transparency, clear accountability and the ability to challenge the result. Not every user needs to understand the mathematical architecture of a model. But people should understand enough to know: what information influenced the decision, what the system was designed to optimise, where uncertainty exists, what kinds of errors are possible, and who remains responsible for the final outcome.
Trust should be calibrated rather than absolute. We should trust systems more in areas where they have demonstrated reliability and less where context, values or rare events matter greatly. We should recognise that AI can be highly capable without being universally competent.
Leaders must also create a culture in which questioning technology is not treated as resistance to innovation. The person who asks, “Why did the system recommend this?” may be protecting the organisation from blind confidence.
The intelligent age will require a new balance. We should not reject systems simply because they are complex. But neither should complexity become an excuse for unquestioned authority. Trust must be earned through performance, explanation and accountability.
We should not trust AI because it speaks with confidence. We should trust it only to the extent that we understand when it deserves to be trusted.