Author entities matter for AI citations because grounded answer engines need to name a source's author with confidence — and a schema.org Person node with a stable @id, real credentials, and sameAs links to authoritative profiles is the cleanest signal that a page has a named human behind it.
Google's Search Quality Rater Guidelines spend most of the E-E-A-T chapter on identifying the person or organization responsible for content and evaluating their expertise. That is a proxy for the same question the LLM has to answer at generation time: "who is this page attributed to and are they credible?"
A useful author entity has five parts: a stable @id anchor (typically https://yoursite.com/#author-name), a real name, a jobTitle, an url pointing to an author page or bio, and a sameAs array linking to LinkedIn, official profiles, and, where they exist, Wikipedia and Wikidata. Reuse the @id from every Article and QAPage the author writes.
Two failure modes to avoid. First, "Admin" or "Editor" bylines with no bio give the engine nothing to work with. Second, declaring an author in schema who is not visibly credited in the DOM violates Google's structured-data guidelines — the visible page must reflect the markup.
Ship one canonical author page per contributor, link out to their real profiles, and reference the same Person @id from every piece they publish.