Public Relations (PR) focuses on relationships with journalists and editors and on earning editorial coverage. Machine Relations (MR) is the Machine Relations discipline for observing and managing how an organization is represented, sourced, and cited across named AI-mediated discovery surfaces. In the MR taxonomy, PR can contribute independent editorial references; MR measures those references separately from identity records, owned content, answer-surface presence, and citation outcomes.
Premise: Public Relations concerns relationships with journalists and editors, editorial coverage, and the narratives carried by that coverage. Machine Relations concerns an organization's observable representation, sourcing, and citation across selected AI-mediated discovery surfaces. This is a Machine Relations category distinction, not a claim about a private decision process used by every AI engine.
Premise: MR does not replace PR and does not treat editorial coverage as a guaranteed cause of AI visibility. It records independent editorial references as one evidence class, then measures identity records, answer-surface appearances, and citations separately.
A 2025 comparative analysis reported a systematic preference for earned-media sources in the AI-search systems and query samples it studied, relative to brand-owned and social sources (Chen et al., 2025).
Premise: MR treats that result as a finding about the study's measured source mix, not as proof that every editorial placement will be retrieved, cited, or attributed to a brand.
| Dimension | Public Relations (PR) | Machine Relations (MR) |
|---|---|---|
| Primary scope | Journalist and editor relationships; editorial coverage | Representation, sourcing, and citation on named AI answer surfaces |
| Observable outputs | Placements, publication details, messages, and media mentions | Entity records, answer appearances, cited sources, and citation rates |
| People or systems observed | Journalists, editors, publications, and audiences | Selected answer engines, their rendered outputs, and cited pages |
| Measurement unit | Coverage and audience metrics defined by the PR program | Repeated queries, answer runs, cited domains, and comparable observation windows |
| Role in the MR taxonomy | Independent editorial references | The broader MR observation and measurement discipline |
Working model (not measured): the Machine Relations Stack organizes practice into earned authority, entity clarity, citation architecture, distribution, and measurement. It places PR-related editorial work within earned authority. The ordering is an MR operating model, not evidence that one layer causes an outcome in another.
Research on 21,143 citations across ChatGPT, Google AI Overview, and Perplexity treated citation selection and citation absorption as separate processes (Zhang, He, and Yao, 2026). In that study, a page could be selected as a citation without contributing equally to the generated answer.
Premise: MR therefore records at least three facts separately:
Premise: MR does not infer causality between those observations. It records a placement without a measured citation as a placement, and it does not treat a citation as proof of brand attribution, recommendation, buyer influence, or a business result.
PR measurement can report whether coverage ran and how the PR program evaluated its reach. MR measurement asks a different question: what did a named answer engine return for a fixed query set during a defined observation window, and which sources did the answer cite?
Pew Research Center found that users in its 2025 Google sample were less likely to click result links when an AI summary appeared, and that clicks on links cited inside the summaries were rare (Pew Research Center, 2025).
Premise: MR measures citation presence separately from referral traffic and does not use citation presence as a proxy for attention, persuasion, or revenue.
Premise: MR uses metrics such as Share of Citation, Citation Velocity, and Entity Resolution Rate as descriptive measurements. These metrics report observed states or changes between comparable windows; they do not identify causality or guarantee an outcome.
Working model (not measured): an organization can use PR to pursue editorial coverage and use MR to evaluate what appears on selected AI answer surfaces. The two practices can share source material while retaining separate objectives and measurement.
A reproducible joint review can record:
Premise: the review does not infer that coverage caused a citation, that a citation caused a recommendation, or that either caused a business result without separate evidence.
Premise: no. PR concerns editorial relationships and coverage. MR observes representation, sourcing, and citation across selected AI-mediated discovery surfaces. In the MR taxonomy, editorial coverage is one possible evidence input rather than the whole discipline.
No. The cited research found earned-media preference in its measured samples.
Premise: MR does not generalize that result into a guarantee that every placement will be retrieved, cited, absorbed into an answer, or attributed to a brand.
Premise: no. MR and PR have different scopes. An organization may pursue editorial coverage through PR and independently measure AI answer-surface representation through MR.
Premise: MR records outcomes against a declared query set, engine set, and observation window. Examples include answer appearances, cited domains, Share of Citation, and Citation Velocity. Those measurements remain separate from impressions, referral traffic, buyer behavior, and revenue.
Premise: yes. MR can inventory owned pages, independent references, entity records, and answer-surface observations regardless of how those assets were created. The measurement does not prescribe one universal acquisition strategy.
Machine Relations' own methodology, dataset, and research pages related to this term. These are self-references, listed separately from Sources — they are not independent evidence.
Supporting research
Framework context