A recent case has brought the limitations of artificial intelligence in official document writing into sharp focus, underscoring that advanced technology cannot replace the need for thorough field research.

As reported by a magazine, a county official recounted his experience using an AI tool to draft a policy-related report. He asked the AI to generate four real case studies that matched the local situation, but the output proved unreliable: two cases were misattributed to the wrong place, and two were completely fabricated. This phenomenon, commonly known as "AI hallucinations," is familiar to many users who rely on such tools for generating content.

Fortunately, this official caught the errors by cross-checking the cases. However, had he submitted the material without verification—perhaps out of convenience or excessive trust in the technology—it could have become the foundation for a new regulation, turning a simple error into a serious governance problem. Recently, another local authority flagged that public sentiment reports were "visibly generated by AI," a trend that merits close attention.

Public sentiment reports fall under the broader category of official documents. Whether AI can be used in their drafting is not merely an ethical question about writing practices. In the opening example, the leader requested detailed information about a specific industry to serve as a basis for policy. Such material cannot simply be "generated" by an algorithm. It requires contacting the relevant department for official data, conducting first-hand visits and investigations, collecting market statistics, and soliciting feedback from enterprises. Relying solely on desk-based information is not an option.

This point is reinforced by the official's experience: he spent an entire night discarding the AI-generated content and starting over, painstakingly verifying every detail against original meeting minutes and enforcement files. He admitted that what was meant to save time almost backfired. This mirrors the experience of many users—revising AI-produced work can take more time and energy than writing from scratch.

The lesson applies well beyond official document writing. Any role or industry that demands accuracy and authenticity should use AI with caution. Especially when it comes to case studies and data, cross-verification is essential. No matter how advanced AI becomes, it cannot fully grasp the nuances of real-world conditions, nor can it replace the value of actual field research in capturing the true pulse of public sentiment.