Studies: 32
Descriptive research records with named source, method or sample context, unit, and limits, counted once per source document. These inform measurement design; they do not establish a universal causal law.
Classified Evidence Library
Machine Relations Evidence Base
A retained evidence library for Machine Relations with source-role classes, counter definitions, and limits. Records inform operating hypotheses; they do not establish a universal causal law.
32 studies·7 independent term adoptions·130 market signals·254 publications cited
What the evidence shows
These records measure or report different samples, source taxonomies, and market behaviors. Muck Rack’s May 2026 edition analyzed more than 25 million cited links from ChatGPT, Claude, and Gemini across 17 industries: 84% fell within its broad earned-media taxonomy, 27% were journalism, and 0.3% were paid or advertorial. Fullintel and UConn separately found that 47% of citations in their health-focused sample came from journalistic sources. These source-composition results do not prove that any source type is the primary causal mechanism for citation.
| Source | Finding | Boundary for Machine Relations |
|---|---|---|
| Muck Rack — May 2026 Generative Pulse | 84% under Muck Rack’s broad earned-media taxonomy; 27% journalism; 0.3% paid or advertorial across 25M+ cited links | Current source-composition point reading; it does not establish placement causality or a primary citation mechanism |
| Fullintel + UConn IPRRC (2026) | 47% of citations in its health-focused sample came from journalistic sources | A separate journalistic-source composition result with a narrower sample and taxonomy |
| Chen et al., ACM / University of Toronto (2025) | Bounded GEO research reports source-composition differences in the authors’ tested prompt and page conditions | Descriptive measurement input; not an earned-media dominance law |
| Aggarwal et al., KDD (2024) | In the authors’ experimental conditions, statistics, quotations, and citations improved generated-answer visibility relative to baselines | Content-feature experiment under stated conditions; not a universal source-preference mechanism |
| Gartner forecast (2024) | Gartner forecast a 25% traditional-search volume decline by 2026 as AI chatbots and virtual agents expand | Forecast about search behavior, not observed replacement evidence |
| OtterlyAI — 1M+ citation dataset (2026) | Descriptive cited-host composition snapshot including Forbes, Reuters, and Wikipedia | Report cited-host mix only; do not convert it into a trust or placement claim |
Counter definitions
Unenriched source references: 91
Retained source pointers awaiting title, author, venue, finding, or statistical enrichment. They remain discoverable but are excluded from the Studies counter.
Retained duplicate study records: 2
Enriched records that resolve to a source document already counted once. They remain retained as provenance but are excluded from the Studies counter.
Independent term adoptions: 7
Verified unaffiliated source documents that use the exact phrase Machine Relations in source-owned text as a present communications framework, are independently resolvable, and are counted once per unique announcement chain. Owned origin records, legacy semantic uses, adjacent context, unenriched pointers, duplicate URLs, and syndications are excluded.
Market signals: 130
Retained market-signal feed records, including market activity and practitioner commentary routed to their own source-role classes. This counter is a retained-record count, not an independence or causal-strength score.
Publications cited: 254
Distinct publication, institution, or source names across the retained evidence files. This is a source inventory count, not an independence or causal-strength score.
Independent term adoption7
Verified unaffiliated source documents that use the exact phrase Machine Relations in source-owned text as a present communications framework, are independently resolvable, and are counted once per unique announcement chain. Owned origin records, legacy semantic uses, adjacent context, unenriched pointers, duplicate URLs, and syndications are excluded.
Mark Huntley and CiteWorks Studio present Machine Relations as an emerging communications discipline for improving the external evidence environment that influences AI-generated answers.
“That is the territory I call Machine Relations.”
A guest practitioner essay makes machine relations a central pillar of modern PR strategy for generative discovery.
“Making machine relations a central pillar of a modern PR strategy is crucial to winning in generative discovery.”
Instant Press defines machine relations as a communications discipline for shaping how AI systems understand and cite a brand.
“Machine relations is the discipline that manages that decision, and most brands do not know it exists yet.”
In a CityBiz Q&A, Carmen Hughes defines Machine Relations as the practice of earning visibility, authority, and citations across AI-powered search platforms.
“Machine Relations is the practice of earning visibility, authority and citations across AI-powered search platforms: ChatGPT, Perplexity, Gemini...”
Ignite X presents machine relations as a parallel PR track serving human audiences and AI systems.
“By “machine relations,” I mean our work now serves a dual function: informing human audiences while also training algorithms.”
Benjamin Chipman reports Gab Ferree's exact-term framing of machine relations as the evolution of media relations for AI-mediated discovery.
“Media relations are becoming machine relations. It's on the comms professionals to learn the patterns of AI and then take action on them.”
An unrelated agency publicly launched a Machine Relations practice and described the term as an emerging discipline, while not naming an originator.
“Machine Relations is now emerging as a new discipline at the intersection of communications, content strategy, and AI.”
Retained adoption-chain duplicates3
Syndicated documents retained for provenance. They use the exact term, but one announcement chain contributes one adoption to the counter.
Ignite X launched a dedicated Machine Relations practice to help brands earn AI search visibility through strategic communications at the intersection of communications, content strategy, and AI.
“Machine Relations is now emerging as a new discipline at the intersection of communications, content strategy, and AI.”
Ignite X differentiates its Machine Relations practice by focusing on PR-rooted authority and credibility signals discussed in AI-search visibility work, rather than SEO tactics.
“Machine Relations is now emerging as a new discipline at the intersection of communications, content strategy, and AI.”
Ignite X has launched a dedicated Machine Relations service to help brands earn AI search visibility and citations on platforms like ChatGPT and Perplexity.
Adjacent market context58
Enriched GEO, AEO, PR, citation, and source-composition context retained for research. These sources do not use the Machine Relations term and do not count as category adoption.
Product Marketing Manager Levi Pillay states that AI systems pick up on consistency over volume, urging brands to focus on a handful of highly credible, relevant sources all reinforcing the same positioning rather than relying on their own site.
“Instead of relying on your own site to do all the work, you want multiple trusted sources reinforcing the same idea about your brand, what you're good at, and when you're the right choice.”
Most people incorrectly equate PR with just getting media coverage, but reality shows that public relations is a broader discipline of developing mutually beneficial relationships beyond just news media.
“Most people think PR = getting media coverage. But here's the reality ✓ Public relations... PR isn't just media relations. The news media is just one of many channels[10].”
Industry debate is increasingly splitting between those who treat AEO/GEO as distinct disciplines and those who say AI search optimization is largely just SEO, with Google representatives quoted as saying the core principles of SEO generally apply and that there is no need for new acronyms.
“You don’t need GEO, LLMO, or anything else.”
Analysis of over one million AI prompts in the May 2026 Generative Pulse report found that 84% of all AI citations originate from earned media sources, including journalism, academic research, and government sources.
“Earned media accounts for 84% of all AI citations in the May 2026 edition, including journalism, academic research, government sources”
Earned media consistently drives 84% of AI citations across ChatGPT, Claude, and Gemini, a structural mechanism documented by four consecutive measurement windows since July 2025.
“Three editions in, the data keeps telling the same story: earned media is what AI trusts”
There is currently no common taxonomy for AI optimization, with agencies and marketers adopting interchangeable acronyms like AEO, GEO, and GSO to describe the same trend of ensuring AI crawlers understand brand information for synthesized answers.
“This is such a new area, that currently there is no common taxonomy. So agencies, publishers, marketers and SEO specialists have adopted a bunch of different acronyms to describe the same trend.”
AEO and GEO are competing names for the same emerging discipline of earning citations inside AI-generated answers, making the naming debate a vendor marketing turf war rather than a real strategic fork.
“AEO versus GEO is a debate about vocabulary, not strategy. Both describe optimizing to be cited and recommended by AI assistants, both build on SEO, and both come down to the same three mechanics: query fan-out, citations, and third-party consensus.”
Fewer than a third of practitioners use GEO, AEO, and LLMO consistently, as the terms are three competing names for a single second discipline—citation optimization—with tactical playbooks that overlap almost completely.
“SEO is one discipline; GEO, AEO, and LLMO are three names for a second discipline, citation optimization, whose tactical playbooks overlap almost completely”
The article frames SEO/AEO/GEO/LLMO/AISO as a naming-fragmentation problem, arguing that the field lacks consensus and that multiple overlapping labels now compete to describe the same AI-search optimization work.
“There is no universally accepted terminology for optimizing visibility in AI-generated results.”
PR is about more than just media relations and involves creating and maintaining an organization's name, crisis preparation and mitigation, and communicating with different interest groups.
“PR is about more than just media relations”
The industry debate over whether SEO, GEO, and AEO are distinct disciplines has yielded a tactical truce, acknowledging that while the acronyms differ, the underlying playbooks for earning AI visibility overlap significantly.
“In the high-stakes arena of digital visibility, a fierce debate rages over whether traditional Search Engine Optimization (SEO) suffices amid the rise of AI-driven answers, or if new disciplines like Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) demand separate playbooks”
Industry coverage notes that the SEO community is actively debating whether AEO and GEO are meaningfully different disciplines or largely new labels for the same AI-search optimization work, highlighting naming fragmentation rather than consensus.
“Whether you call it SEO, AEO, GEO, or AI SEO, the underlying goal is the same: get your content cited, surfaced, and trusted by AI systems.”
Unenriched adoption source references140
Retained source pointers awaiting source-owned text enrichment. No category-adoption or causal inference is made, and future feed records default to excluded.
Descriptive research32
Peer-reviewed and industry research on AI citation source composition, retrieval, and visibility. Each record retains its source, sample, and causal limits.
In a health-focused AI-search sample, 47% of citations came from journalistic sources.
47% journalistic sources in the study sample — Fullintel/UConn IPRRC
Bounded GEO study reporting source-composition differences between earned, brand-owned, and social sources in the authors’ tested prompt/page conditions.
Primary paper: arXiv 2509.08919 / ACM-affiliated GEO work; units and engine/query conditions must stay attached to the study.
In the authors’ experimental GEO conditions, adding statistics, quotations, and citations improved generated answer visibility relative to baselines.
Primary paper: arXiv 2311.09735 / KDD 2024; report effect sizes with the study denominator and method.
The May 2026 edition analyzed more than 25 million cited links from ChatGPT, Claude, and Gemini responses across 17 industries. Under Muck Rack's broad taxonomy, 84% were earned media, 27% were journalism, and 0.3% were paid or advertorial.
May 2026: 84% broad earned-media taxonomy; 27% journalism; 0.3% paid/advertorial (25M+ cited links)
Independent third-party evidence from earned media accounts for 85.5% of AI citations, while brand-owned websites contribute minimally to AI engine recommendations.
85.5%
5W's April 2026 study quantifies that 85.5% of AI citations reference earned media sources, establishing it as the primary input for brand visibility in AI-mediated discovery.
85.5%
OtterlyAI reported cited-host composition from a 1M+ citation dataset, including frequent appearances by Forbes, Reuters, and Wikipedia.
73% of sites block AI crawlers — invisible to citation systems by default
Muck Rack Generative Pulse reports cited-link composition by edition and taxonomy, including high non-paid and earned-media shares in AI-cited links.
Generative Pulse cited-link composition; edition and taxonomy limits required before comparing shares
Signal Genesys reports an October 1-December 24, 2025 domain-level analysis of 179.5 million citation records across 6.1 million unique domains and six LLM platforms, with platform-specific coverage, volume, rank, citation-score, and share-of-voice measures.
179.5M citation records; 6.1M unique domains; six LLM platforms; Oct. 1-Dec. 24, 2025 domain-level analysis
17.2 million distinct AI citations analyzed across Q4 2025. No single optimization strategy works across ChatGPT, Gemini, Perplexity, Claude. Model-specific citation behavior requires multi-surface approach.
17.2M citations analyzed: no single tactic dominates all AI platforms
Muck Rack Generative Pulse reported that 82% of AI-cited links in its December 2025 point reading fell under its earned-media taxonomy.
December 2025 point reading: 82% of AI-cited links under Muck Rack earned-media taxonomy
Columbia Journalism Review Tow Center tested eight generative-search tools with a 1,600-query design over publisher/article samples, finding frequent citation absence and publisher/URL attribution errors.
1,600 queries across eight AI search tools; 49% of answers contained citations; 31% of cited links pointed to the original publisher
Retained source references93
Unenriched source pointers and duplicate study records retained for provenance and later enrichment. They are excluded from the Studies counter and use URL-derived labels instead of blank headings.
A Generative Engine Optimization study of 87 stories across 8 AI platforms found that distributing content through earned media channels delivers a 239% median lift in AI search visibility compared to brand-owned content alone.
239% median lift
The February 2026 Fullintel-UConn study reported that 47% of citations in its health-focused AI-search sample came from journalistic sources.
47% journalistic sources in the study sample
Practitioner and industry commentary5
Practitioner columns, contributed council posts, vendor essays, and industry commentary. These are commentary signals rather than independent empirical proof.
Forbes Business Council piece argues AI summaries have made PR more important than ever — 'Public relations (PR), long treated as a parallel track to Search, is suddenly having an outsized influence on the answers people see.'
“Public relations (PR), long treated as a parallel track to Search, is suddenly having an outsized influence on the answers people see. Mentions in respected outlets, expert commentary... [are what shape AI answers].”
PR now has outsized influence on AI answers — Forbes Business Council, January 2026
Forbes Agency Council piece on 'Using Public Relations to Support Your GEO Strategy' opens by stating GEO success starts with earned media and cites Muck Rack: 89% of all AI-cited links are earned media.
“Muck Rack analyzed over one million links cited by ChatGPT and found that 89% of citations originated from earned media... Companies proactively engaged in PR efforts are reaping a new advantage: visibility within generative AI engines.”
89% of AI citations from earned media — Muck Rack (1M+ citation dataset), cited in Forbes
Deloitte's GEO strategy guide for CMOs identifies earned media and thought leadership as a primary tactic: 'build brand equity through earned media and thought leadership' alongside technical GEO tactics.
“Manage reputation proactively. Marketers should update brand information in industry databases, encourage customers to evaluate products on review websites, and look for opportunities to build brand equity through earned media and thought leadership.”
Deloitte GEO strategy for CMOs: earned media is a primary AI visibility tactic
Peer-reviewed GEO research finds AI search engines show 'systematic and overwhelming bias towards Earned media (third-party, authoritative sources) over Brand-owned and Social content.' Strategic recommendation: 'dominate earned media to build AI-perceived authority.'
“AI Search exhibit a systematic and overwhelming bias towards Earned media (third-party, authoritative sources) over Brand-owned and Social content, a stark contrast to Google's more balanced mix.”
AI search shows systematic bias toward earned media over brand-owned content — Chen et al. ACM 2025
17 PR and marketing executives advising on GEO strategy unanimously center their guidance on earned media: 'success with GEO depends on creating content that resonates with both audiences and algorithms — and it starts with credibility built through earned media.'
“Success with GEO depends on creating content that resonates with both audiences and algorithms — and it starts with credibility built through earned media. Mentions in reputable outlets that are structured, quotable, and clearly attributed have a higher chance of appearing in AI snippets.”
17 PR agency executives: GEO success 'starts with credibility built through earned media' — Forbes 2025
Market signals125
PR firm reorganizations, analyst forecasts, product launches, and adjacent market activity around AI search visibility. Practitioner commentary is classified separately, and these signals do not count as independent term adoption.
Ignite X announced a dedicated Machine Relations practice via GlobeNewswire, positioning Machine Relations as an emerging discipline in AI search visibility.
“Machine Relations is now emerging as a new discipline at the intersection of communications, content strategy, and AI.”
Launch announced April 9, 2026
The PR industry uses key terms like 'Earned Media' for publicity gained through PR efforts and 'Media Relations' for mutually beneficial relationships with media, excluding paid advertising.
null
Nine experts failed to agree on whether to call the discipline GEO, AEO, AI SEO, or Relevance Engineering, with one closing the debate only by resetting the definition.
9
Branded search dependency has dropped 6.7 percentage points from 74.7% two years ago, indicating consumers increasingly discover products via conversational AI queries rather than specific brand name searches.
6.7 percentage point drop
Jaxon Parrott's Entrepreneur article arguing that public relations has become Machine Relations was syndicated to Yahoo Finance and MSN, then amplified by Entrepreneur on X.
Article live on Entrepreneur, syndicated by Yahoo Finance and MSN, and shared by Entrepreneur's official X account on May 6, 2026
A May 2026 Generative Pulse report is reported to show that earned media accounts for 84% of AI citations across ChatGPT, Claude, and Gemini, indicating that third-party editorial coverage is the dominant source layer in AI answers.
84%
Brand search volume has a correlation of 0.334 with AI citations, the strongest factor measured, while earned media remains essential for authority and discoverability
0.334
Press release syndicated to Yahoo Finance, Business Insider Markets, Benzinga, Barchart, and GlobeNewswire defining Machine Relations as the discipline where GEO, AEO, SEO, and PR converge in AI search. Appeared in Google AI Overview within hours of publication.
5 syndication nodes (DA 62-95) live within hours; appeared in Google AI Overview same day
AI-driven discovery restores PR's strategic role. PR outputs — media coverage, expert commentary, institutional evidence — are exactly what AI systems prioritize.
Gartner: PR spend projected to double by 2027 driven by AI discovery needs
Academic paper published on arXiv exploring machine relations era dynamics in AI brand discovery.
Academic paper documenting MR dynamics in AI brand discovery
Earned media's role in AI search results is growing dramatically, with 85% of GEO results from earned sources and the figure increasing.
85% of GEO results from earned media
Analysis of 23,000+ AI interactions revealed that earned media sources account for nearly half (48%) of all citations in branded LLM queries, with LLMs synthesizing signals from across the ecosystem rather than just pulling from the brand website.
48%
Legacy and adjacent semantic occupation10
Human-machine, robotics, manufacturing, ethics, and machine-behavior uses retained for disambiguation research only. They do not count as adoption of the 2024 Machine Relations communications discipline.
Advances in AI and robotics prompt discussions on human-machine relations, emphasizing ethical values and ontological distinctions to guide interactions.
“Development of sophisticated AI and robotics technologies has motivated claims for their ontological and value parity with humans.”
Machine Behavior is proposed as a new academic discipline focused on the scientific study of intelligent machines' behavior as actors, distinct from computer science.
“We advocate the need for a new, distinct scientific discipline of Machine Behavior: the scientific study of behavior exhibited by intelligent machines.”
MIT Media Lab researchers’ idea of a new discipline for studying machine behavior shows that naming machine-centered fields has precedent, but this source uses "machine behavior," not "machine relations."
“The new field of machine behavior”
Academic literature explores the evolving dynamics of **human-machine relations** as a distributed agency phenomenon.
“This article delves into the intricate and evolving dynamics of human–machine relations.”
Relational autonomy can incorporate human-machine relations through decision-theoretic frameworks, enhancing human decision-making while setting boundaries to preserve autonomy.
“relational autonomy account and describe how it amounts to a distinct way of understanding the decision-making complexities of embedded social beings.”
Organizations are increasingly recognizing the strategic importance of defining human-machine relationships, with 43% of surveyed organizations already appointing dedicated leaders to manage human-machine collaboration.
“defining the human-machine relationship is not an abstract academic exercise. There are real-world effects now of how we will make decisions, how and what kinds of work will be performed, how we will organize our companies.”
Human-Robot Interaction (HRI) is established as a new multi-disciplinary field addressing cognitive interactions and teaming between humans and machines, distinct from broader human-system interaction.
“Human-robot interaction (HRI) is a relatively new, multi-disciplinary field that addresses how people work or play with robots versus computers or tools.”
Recent peer-reviewed work on reciprocal learning in human–machine collaboration indicates continued academic recognition that machine systems have their own learning and interaction dynamics, lending support to the debate that GEO, AEO, and AI SEO are fragments of a larger machine-facing discipline.
The evolution from automation to autonomy in AI is reshaping human-machine relations in diverse contexts.
MIT Media Lab’s proposal to treat machine behavior as a discipline shows that adjacent terminology in the AI ecosystem is being formalized around how machines act, decide, and are studied, which supports the emergence of category language like 'machine relations' even though the exact MR term is not used.
Origin and provenance3
Jaxon Parrott and AuthorityTech records supporting the public origin chain. They preserve provenance but do not count as independent term adoption.
Parrott’s 2026 framing positions Machine Relations as the parent discipline for AI citation and recommendation work, with GEO and AEO treated as execution layers rather than substitutes.
“Machine Relations is the parent category.”
AuthorityTech’s public framework explicitly states that the AI visibility industry is experiencing naming fragmentation and positions Machine Relations as a unified discipline that contains GEO and AEO as tactics rather than standalone categories.
“The AI visibility industry is currently experiencing a naming fragmentation identical to what SEO experienced in the early 2000s.”
AuthorityTech’s March 2026 release formalized Machine Relations as the discipline of earning AI citations and recommendations by making a brand legible, retrievable, and credible inside AI-driven discovery.
“Machine Relations (MR) — the discipline of earning AI citations and recommendations for a brand by making that brand legible, retrievable, and credible inside AI-driven discovery.”