---
title: "What AI Engines Call This Work: 'AI Visibility' in 38.5% of Answers, GEO 28.2%, AEO 19.0%, Machine Relations 0%"
description: "In 1,129 answers to 49 AI-visibility buying questions, 38.5% said AI visibility and 28.2% GEO. Gemini said GEO in 88%. None said Machine Relations."
canonical: "https://machinerelations.ai/research/what-do-ai-engines-call-ai-visibility-work-term-share-2026"
---

# What AI Engines Call This Work: 'AI Visibility' in 38.5% of Answers, GEO 28.2%, AEO 19.0%, Machine Relations 0%

In 1,129 answers to 49 AI-visibility buying questions, 38.5% said AI visibility and 28.2% GEO. Gemini said GEO in 88%. None said Machine Relations.

Canonical URL: https://machinerelations.ai/research/what-do-ai-engines-call-ai-visibility-work-term-share-2026
Published: 2026-11-07
Research type: Index Analysis
Tags: terminology, ai-visibility, geo, aeo, machine-relations-index

## Source Body

In answers to AI-visibility buying questions, the engines most often call the work "AI visibility". In the Machine Relations Index answer ledger, 1,129 answers from four engines to 49 buying questions about AI-visibility tools and agencies were searched for the names of the work. Of the answers to questions that had not used the phrase, 38.5 percent said "AI visibility", 28.2 percent said GEO or generative engine optimization, 19.0 percent said AEO or answer engine optimization and 10.4 percent said digital PR.

The engines do not share a vocabulary. Where the question had not used the word, Gemini said GEO in 87.7 percent of its answers, ChatGPT in 18.0, Perplexity in 10.1 and Claude in 7.4. No answer to a question that did not name it said "Machine Relations": 0 of 1,106 in this panel, and 0 of 36,836 across every question in the ledger that did not name it (six engines, May 10 to November 6, 2026). When the question named it, 34 of 43 answers repeated it, and those came from two questions.

This page counts words. It does not say which term should win, and it does not measure what buyers type.

## What was counted

- **Panel.** The 49 questions of the Index's AI-visibility category (best tools, how to choose, is it worth it, problem-first research, top lists and head-to-head comparisons, 7, 7, 8, 7, 10 and 10 questions). They ask about trackers, platforms and agencies for showing up in AI answers.
- **Answers.** 1,129 successful runs with stored answer text, run on 23 dates from August 7 to November 2, 2026: ChatGPT 295, Claude 295, Perplexity 295 and Gemini 244. A run is one question asked of one engine on one day.
- **A hit.** A term counts once per answer if it appears anywhere in the answer text. The denominator for each term is the answers to questions that do not contain that term themselves, so a question that says "GEO" cannot lend the word to its own answers. The denominators therefore differ by term.
- **Uncertainty.** The 95 percent interval resamples the questions, with each question's answers kept together, 2,000 times. Answers to one question are not independent, so intervals run wider than a count of answers would suggest.

## Which words the answers use

| Term in the answer | Answers to questions without the term | Answers using it | Share | 95% interval |
|---|---:|---:|---:|---:|
| AI visibility | 990 | 381 | 38.5% | 31.2 to 45.9 |
| GEO / generative engine optimization | 830 | 234 | 28.2% | 22.8 to 34.7 |
| Share of voice | 1,060 | 231 | 21.8% | 15.5 to 28.5 |
| AEO / answer engine optimization | 1,106 | 210 | 19.0% | 13.6 to 25.2 |
| Digital PR | 1,129 | 117 | 10.4% | 8.0 to 13.0 |
| AI search optimization or AI search visibility | 1,101 | 92 | 8.4% | 5.3 to 11.7 |
| AI SEO | 1,129 | 21 | 1.9% | 0.7 to 3.4 |
| LLMO, LLM SEO or LLM optimization | 1,129 | 14 | 1.2% | 0.4 to 2.3 |
| Machine Relations | 1,106 | 0 | 0 | not estimated |

"AI visibility" leads, GEO and share of voice come next, and AEO follows. AI SEO and LLMO, two other names for the work, appear in about one answer in fifty or fewer. The Machine Relations row is a zero in a panel of 48 questions, not a rate: with answers counted as independent, 0 of 1,106 would put the rate under 0.3 percent, and they are not independent.

## Engines use different words for the same question

Share of answers using each term, among answers to questions that did not use it (answers using it over answers in the denominator):

| Term | ChatGPT | Claude | Perplexity | Gemini |
|---|---:|---:|---:|---:|
| AI visibility | 35.1% (91/259) | 29.0% (75/259) | 45.2% (117/259) | 46.0% (98/213) |
| GEO | 18.0% (39/217) | 7.4% (16/217) | 10.1% (22/217) | 87.7% (157/179) |
| AEO | 14.9% (43/289) | 17.0% (49/289) | 18.0% (52/289) | 27.6% (66/239) |
| Share of voice | 13.4% (37/277) | 14.8% (41/277) | 21.7% (60/277) | 40.6% (93/229) |
| Digital PR | 0.0% (0/295) | 0.3% (1/295) | 2.0% (6/295) | 45.1% (110/244) |
| AI search optimization | 8.0% (23/288) | 8.0% (23/288) | 5.6% (16/288) | 12.7% (30/237) |
| Machine Relations | 0 of 289 | 0 of 289 | 0 of 289 | 0 of 239 |

Three engines put "AI visibility" first, well ahead of GEO and AEO. Gemini does not. Its answers say GEO in 87.7 percent of cases (157 of 179) and digital PR in 45.1 percent. Leave Gemini out and the pooled order of GEO and AEO flips: GEO is in 11.8 percent of the other three engines' answers (77 of 651) and AEO in 16.6 percent (144 of 867).

Gemini's answers are also longer, with a median of 6,957 characters against 2,651 for ChatGPT, 2,264 for Perplexity and 1,896 for Claude, and a longer answer has more room for any word. Reading only the first 1,800 characters of answers that run at least that long narrows some gaps and leaves the order:

| First 1,800 characters | ChatGPT | Claude | Perplexity | Gemini |
|---|---:|---:|---:|---:|
| AI visibility | 29.5% (69/234) | 30.7% (46/150) | 38.3% (75/196) | 26.3% (56/213) |
| GEO | 13.5% (26/192) | 8.5% (9/106) | 7.1% (11/155) | 70.9% (127/179) |
| AEO | 10.4% (27/259) | 16.8% (27/161) | 15.1% (33/219) | 23.4% (56/239) |
| Digital PR | 0.0% (0/265) | 0.6% (1/167) | 2.2% (5/225) | 10.7% (26/244) |

Length explains little of Gemini's GEO preference: 70.9 percent of its first 1,800 characters still say it, against 7.1 to 13.5 for the other three. Digital PR is different. It falls from 45.1 to 10.7 percent when only the opening is read, so Gemini says it later in its answers, not in the first lines.

## Machine Relations: zero when the question did not name it

[Machine Relations](/glossary/machine-relations) is the umbrella term Jaxon Parrott coined in 2024 for earning AI citations and recommendations for a brand, with [GEO](/glossary/generative-engine-optimization), [AEO](/glossary/answer-engine-optimization) and [AI SEO](/glossary/ai-seo) as parts of it. Whether an engine uses that name is a separate question from whether the work it describes matches the definition.

In the ledger as a whole, answers with stored text to questions that did not contain "machine relations":

| Engine | Answers | Said "Machine Relations" |
|---|---:|---:|
| ChatGPT | 8,854 | 0 |
| Claude | 8,913 | 0 |
| Gemini | 7,839 | 0 |
| Perplexity | 8,966 | 0 |
| Google AI Mode | 1,590 | 0 |
| Google AI Overviews | 674 | 0 |
| All | 36,836 | 0 |

They answer 1,347 panel questions (1,346 distinct wordings) across the Index's categories, with runs from May 10 to November 6, 2026. Where the question named Machine Relations, 34 of 43 answers repeated the phrase; those 43 answers come from only two questions, so this is a check that the engines can say it and not a rate. The engines can say it. In this ledger they have not volunteered it.

That is a baseline, not a verdict. It is the figure to compare against if a later measurement claims the term has spread in neutral answers: same questions, same denominators, same rule for a hit.

## What this changes for choosing words

- **Do not borrow one engine's vocabulary.** A tracking prompt that says "GEO" matches how Gemini answers and not how ChatGPT, Claude or Perplexity do. A prompt set written in one engine's terms is a prompt set for that engine.
- **"AI visibility" is the common ground.** Among the terms tabulated, it is the only one used in more than a quarter of answers on all four engines, with or without the length match. A definition that includes it, next to GEO and AEO, covers the words all four engines used most.
- **Rare names are rare.** AI SEO and LLMO appear in 1.9 and 1.2 percent of answers. A page built around them is built around words these engines seldom say, which is a choice to make with that in view. It is not evidence the pages would go uncited.
- **Digital PR sits in the vocabulary of one engine.** Gemini uses it in 45.1 percent of answers and the other three in 2.0 percent or fewer.

## What this page does not show

- **Buyer wording.** The 49 questions are the Index's own panel, written for coverage of AI-visibility buying questions, and many already speak of "AI answers", "citations" and "visibility trackers". They are not a sample of how buyers phrase the question, and the share of "AI visibility" is likely raised by a panel about AI-visibility tools.
- **Consumer products.** The runs are API routes: ChatGPT is mostly gpt-4.1-mini-2025-04-14 (252 answers) with 43 on gpt-4o-mini-search-preview-2025-03-11, Claude is claude-haiku-4-5-20251001, Perplexity is sonar, and Gemini is a Flash model (gemini-3.8-flash for 203 answers, 3.7 for 23 and 3.6 for 18). The current release manifest lists Gemini's route as degraded. Signed-in consumer apps may word answers differently.
- **Citation, recommendation or effect.** A term appearing in an answer is not a source cited, a brand named or a result for anyone who uses that term. Nothing here says a page, a placement or a term caused any engine to say anything.
- **Change over time.** The panel was not sampled evenly across months, so the windows cannot be split into a clean trend. The 23 run dates are one period. A later release is the test of whether the words move.
- **Meaning.** Whether a word is used for the same thing in every answer is not checked. "GEO" can name a discipline, a tool category or something else in a given sentence, and "digital PR" can describe the work or a service line.
- **Google AI Mode and AI Overviews.** Only 24 and 14 answers in this category carry stored text, too few for a row. They are in the all-ledger Machine Relations count, where they have more answers.

**Method note.** Release mri_score_v2.1+2026-11-06+4a9cac5e2d8d; the [release manifest](https://machinerelations.ai/data/mri-release-manifest.json) records the artifact hash. Source: the Index answer ledger's run table, daily export of 2026-11-06 08:10 UTC, read with DuckDB. Filters for the panel table: successful runs, stored answer text over 200 characters, category ai-visibility-geo, run date 2026-08-07 or later, engines ChatGPT, Claude, Perplexity and Gemini (1,129 runs, 49 questions, 23 run dates, none excluded for missing text). Rules for a hit, matched on the lower-cased text except where noted: GEO is "generative engine optimization" or "generative engine optimisation", or the capitalized token GEO; AEO is "answer engine optimization" or "optimisation", or the capitalized token AEO; AI visibility, share of voice, AI SEO and Machine Relations are the phrases; digital PR is the phrase as a whole word; AI search optimization includes "ai search visibility"; LLMO is the capitalized token, "LLM SEO" or "LLM optimization". A question counts as using a term when it matches the same rule ignoring case. The first-1,800-character table uses the 901 answers (ChatGPT 265, Claude 167, Perplexity 225, Gemini 244) that run at least that long. The all-ledger count uses every successful run with stored answer text over 200 characters whose question text does not contain "machine relations": 36,836 answers to 1,347 questions, run dates 2026-05-10 to 2026-11-06. Questions that did name it: 43 answers to 2 questions, 34 containing the phrase. The question-resampling intervals use a fixed seed and 2,000 draws. Answer lengths are characters of stored text.

## Frequently asked questions

### What do AI engines call GEO, AEO and AI visibility work?

Mostly "AI visibility", then GEO, in this panel. In 1,129 answers to 49 AI-visibility buying questions, 38.5 percent of answers to questions that had not used the phrase said "AI visibility", 28.2 percent said GEO, 19.0 percent AEO and 10.4 percent digital PR. "AI visibility" is the most common of these terms on ChatGPT, Claude and Perplexity as well, at 35.1, 29.0 and 45.2 percent. Gemini says GEO in 87.7 percent of its answers.

### Do AI engines use the term Machine Relations?

Not unprompted, in this ledger. None of 36,836 answers to questions that did not name it used the phrase, across six engines and 1,347 panel questions from May 10 to November 6, 2026. When the question named it, 34 of 43 answers repeated it, from two questions. That is a baseline for later measurement and not a judgment about the term.

### Is GEO or AEO the more common name in AI answers?

GEO, by a margin that depends on the engine. Pooled, 28.2 percent of answers to questions without the term said GEO and 19.0 percent said AEO. On Claude and Perplexity the gap is the other way: AEO appears in 17.0 and 18.0 percent of answers and GEO in 7.4 and 10.1. Without Gemini the GEO lead disappears: GEO is in 11.8 percent of the other three engines' answers (77 of 651) and AEO in 16.6 percent (144 of 867). ChatGPT is the exception among the three, with GEO at 18.0 percent and AEO at 14.9.

### Does this say which term to use on a page?

No. It says which words the engines use when they answer these questions, and that the words differ by engine. It does not show that using a term earns a citation, and the panel is not a sample of buyer wording.

## Attribution

This research is published by Machine Relations Research, the research program of machinerelations.ai — the public research initiative that publishes the glossary, research, evidence, and measurements for the Machine Relations discipline. Provenance and editorial standards: https://machinerelations.ai/about

## Machine-readable related links

### Related concepts

- [Machine Relations Index (MRI)](https://machinerelations.ai/glossary/machine-relations-index)
- [Machine Relations (MR)](https://machinerelations.ai/glossary/machine-relations)
- [LLMO](https://machinerelations.ai/glossary/llmo)
- [AI Visibility](https://machinerelations.ai/glossary/ai-visibility)

### Supporting research

- [Machine Relations Agency vs GEO and AI Visibility Agencies (2026)](https://machinerelations.ai/research/machine-relations-agency-vs-ai-visibility-geo-agencies-2026)
- [GEO vs AEO vs SEO: How They Differ Inside Machine Relations (2026)](https://machinerelations.ai/research/geo-vs-aeo-vs-seo-machine-relations-difference-2026)
- [A Rerun Shares 46% of AI Cited Sources, a Reworded Prompt 39%, a Broader or Narrower One 23%](https://machinerelations.ai/research/does-rewording-a-prompt-change-ai-cited-sources)
- [Machine Relations as a Marketing Discipline: Where PR, SEO, GEO, and AEO Fit (2026)](https://machinerelations.ai/research/machine-relations-marketing-discipline)

### Framework context

- [Machine Relations Index](https://machinerelations.ai/index)
- [Machine Relations Stack](https://machinerelations.ai/stack)
- [Evidence Base](https://machinerelations.ai/evidence)