Field note

AI Does Not Just Misidentify People. It Synthesizes Them.

What same-name search errors reveal about the fragility of human data infrastructure.

Region

Human data infrastructure / AI

Related state
Synthesized

When AI searches for a person, the problem is not only a wrong answer.

People with the same name. Companies in similar industries. Images that appear beside each other in search results. Past job titles. Social profiles. Fragments from earlier conversations. Company names, roles, industry keywords.

When these fragments are placed near each other, AI may connect them as information about the same person.

One person's profile may merge with another's. An executive record from one company may attach to someone else's public image. Industry knowledge from a conversation may be written as if it were the person's own career. And AI may narrate the result in calm, plausible prose.

This is not only hallucination. It is mislinking inside human data infrastructure.

People do not exist as search fragments. Yet inside sales databases, recruiting databases, CRMs, and enrichment services, people are handled as bundles of fragments: name, face, company, title, email, social presence, past employers, industry, touchpoints, inferred interest, contactability.

When those fragments are bound together, the person becomes not a self, but usable person data.

AI adds another movement. The data is explained. Summarized. Given meaning. Spoken as biography. Rendered as personhood.

The danger is not only that AI errs. The danger is that AI generates a person who was never there.

In search results, a person may become someone else. In a sales database, a person becomes a contactable asset. In recruiting infrastructure, a candidate. In a CRM, a contact. In AI prose, a plausible story.

This conversion usually happens where the person cannot see it.

The problem of human data infrastructure is not only that fragments are collected. It is that collected fragments can be wrongly joined — and spoken as if they were one life.

AI may appear to be researching a person. Sometimes it is generating one.

Observing that difference will matter.

← All notes