Hong Kong Feng Shui Practitioners Test AI: How Much Should We Trust a Chatbot’s Reading?
In February 2026, the South China Morning Post interviewed Hong Kong feng shui practitioners about people asking chatbots for fortunes and feng shui advice, and whether AI could replace a traditional consultation. Practitioners warned that a model might misuse classical texts and terminology or offer unsuitable predictions. The debate is less a contest over who is more mystical than a product question: what sources does AI use when it handles a tradition, and how can an error be found?
If a user uploads a room photo, a model may describe visible doors, windows, furniture and light. But moving from image recognition to a feng shui interpretation involves traditional rules, differences between schools and the user’s aims. Describing a room does not show that a system understands the full tradition.
What the report’s interviewees said
The South China Morning Post’s February 17 report recorded Hong Kong practitioners questioning chatbots’ interpretive ability and noted that residents were asking them about fortunes. The practitioners said AI trained on available data might make mistakes with metaphysical concepts. These are interviewees’ views and examples, not a uniform test of every AI system.
Before asking whether AI “can do feng shui,” define the task. Is it identifying which way a door faces in a photo, explaining a Bagua term or offering spatial advice from a particular school? Without a clear task and evaluation criteria, the broad question cannot be checked.
Calculation, knowledge and on-site judgment
Digital tools can process maps, directions or user inputs and retrieve traditional concepts from a knowledge base. A language model can organize the results into conversation. If those steps are not separated, a model may combine plausible-sounding claims into a fluent answer. A response that sounds reasonable may still misrepresent a school of thought.
A traditional consultation may involve observing the site, considering its history and asking about the client’s needs. An AI product should say what information it can access: does it know the actual orientation, building structure and surrounding environment? A single interior photo cannot supply details the user has not provided.
Let users inspect the basis of an interpretation
A useful AI feng shui experience should identify the terminology and rules it uses, say which inputs came from the user and which observations came from image recognition, and allow users to correct errors. For renovation safety, structural changes or health questions, it should explain that traditional interpretation cannot replace professional advice from engineers, building experts or clinicians.
Mystilink can show its value through a clear workflow: visible knowledge sources, spatial inputs, rule explanations and personalized conversation. If a feature has not launched, it should be described as a plan rather than presented as a current capability.
Beyond the human-versus-machine debate, ask what users need
Some users want an explanation of traditional knowledge; some want to discuss a home layout; others are looking for reassurance. AI does not have to replace every practitioner, and human service does not have to be called obsolete. Each offers different experience and responsibility. Users should know which one they are choosing.
Trust comes from boundaries people can check
The Hong Kong discussion is a reminder that fluent language and cultural understanding are not the same thing. AI can help organize information, while users should be able to see its basis, limits and data practices. Mystilink’s distinction from other products should show up in those design choices, not in an unsupported claim that it is “more accurate than a human.”