AI search has spawned a pile of acronyms, most of them used loosely. This glossary defines the terms that actually matter, plainly, with links to deeper guides where we have them. It is meant as a reference: skim the list, jump to a term, or read it through. Where a term is contested or overhyped, we say so.
AEO
Answer Engine Optimization. The practice of structuring content so an AI or search engine can lift a clean, direct answer out of it. AEO focuses on the on-page side: direct answers near the top, question-style headings, self-contained passages. See our full guide to answer engine optimization.
AI Mode
Google's conversational, generative search experience, which passed 1 billion monthly users in its first year (announced at Google I/O 2026). Launched in 2025, it lets users ask follow-up questions and receive fully AI-generated answers rather than a list of links. It sits alongside, not instead of, the classic results page.
AI Overview (AIO)
The AI-generated summary Google displays above the traditional search results for many queries. AI Overviews now appear on roughly 25–50% of US searches depending on the keyword set measured — BrightEdge's commercial-keyword tracker reports ~48%; broader panels such as Semrush and Conductor report 25–30% — and when present, reduce click-through to the pages below by roughly 35–60%, depending on the study (Ahrefs, Seer Interactive). Note that "AIO" is sometimes also used to mean "AI Optimization," which is a source of confusion; in this glossary AIO means AI Overview.
AI SEO
A loose umbrella term for optimizing to appear in AI-driven search. In practice it means the same thing as AEO, GEO and LLMO combined, and different people use it to mean different things. Treat it as a general label, not a precise discipline.
AI Visibility
How often, and how accurately, your brand appears inside AI-generated answers across platforms like ChatGPT, Gemini, Perplexity and Google's AI surfaces. It is the umbrella outcome that AEO, GEO and LLMO all work toward. See our guide to AI visibility.
Answer Engine
A system that answers a question directly rather than returning a list of links to choose from. Featured snippets were an early form; AI Overviews, ChatGPT and Perplexity are the current ones. The shift from search engines to answer engines is the reason terms like AEO exist.
Chunking
Breaking content into small, discrete pieces on the theory that this makes it easier for AI to extract. It is frequently sold as an AI-optimization tactic, but Google's 2026 guidance states plainly that you do not need to chunk content for its generative features, since its systems can understand multiple topics on a single page.
Citation
A reference an AI answer makes to a source. Being cited means your content or brand is one of the sources the model drew on and named when forming its answer. In AI search, citations play the role rankings played in traditional search: they are how you get credit for being an answer.
Citation Rate
A measurement of how often your brand or site is cited across a defined set of AI answers or prompts. Along with share of voice and prominence, it is one of the core metrics for tracking AI visibility, and it is more meaningful than raw rankings, because it measures inclusion in answers rather than position in a list.
Entity
A distinct, identifiable thing, a brand, person, product or place, that a model or search engine recognizes as a single, unambiguous node. Entity clarity is central to AI visibility: a model can only confidently recommend a brand it can cleanly identify. Ambiguous names and inconsistent descriptions create entity confusion, where the model cannot tell who you are or mixes you up with something else.
GEO
Generative Engine Optimization. The practice of building the off-site authority and signals, mentions, reviews, coverage, that make a generative model confident enough to cite and recommend your brand. Unlike the other acronyms, GEO has an academic origin: a 2023 research paper by Pranjal Aggarwal et al. at IIT Delhi and Princeton (published at KDD 2024) that introduced the term and a benchmark. See our guide to generative engine optimization.
Generative Engine
An AI system that generates a synthesized answer from multiple sources, rather than retrieving and ranking documents. ChatGPT, Perplexity, Gemini and Google's AI Overviews are all generative engines. The term comes from the GEO research and is the "engine" that GEO optimizes for.
Grounding
Basing a model's generated answer on retrieved, external source material so that the answer is accurate and attributable, rather than produced from the model's training alone. A "grounded" answer is one anchored to sources it can cite. Grounding is what makes AI search answers verifiable, and it is why being present in the retrieved sources matters so much.
Hallucination
When a large language model produces false or fabricated information stated with confidence. Hallucinations are why accuracy and consistent, verifiable information about your brand matter: a model working from thin or contradictory information about you is more likely to describe you wrongly.
Inference
The stage at which a trained model generates an output in response to an input, as opposed to training, when it learns. The distinction matters for AI visibility because a model can encounter your brand at two moments: during training (what it already knows) and at inference (what it retrieves live to answer a specific question). Optimizing for both is the idea behind LLMO.
Knowledge Graph
A structured database of entities and the relationships between them. Google's Knowledge Graph is the best-known example, powering the info panels in search. Models and search engines use entity graphs to understand what a brand is and how it relates to a category, which is why consistent, structured facts about your brand help you get recognized correctly.
LLM
Large Language Model. An AI model trained on very large amounts of text to understand and generate language, ChatGPT, Gemini, Claude and Grok are all built on LLMs. LLMs power the answer engines that AI search runs on, so understanding roughly how they retrieve and generate is the foundation of optimizing for them.
LLMO
Large Language Model Optimization. Getting large language models to understand your brand correctly and surface it in their answers. In practice it overlaps almost entirely with AEO and GEO; its one distinct emphasis is the model and entity layer, whether the model knows who you are, not just whether it cites you live. See our guide to LLMO.
llms.txt
A proposed plain-text file, similar in spirit to robots.txt, meant to guide AI models to a site's key content. It is an unofficial, emerging convention with no confirmed adoption by major platforms, and Google has stated it does not use these files and that creating one neither helps nor harms visibility in Google Search. Treat it as a low-cost experiment, not a meaningful lever.
Prominence
How high or early your brand appears within an AI answer, named first versus buried at the bottom of a list. Prominence matters because presence alone is misleading: being mentioned sixth in a list is close to invisible. Good measurement tracks prominence, not just whether you appear at all.
Prompt
The input or question a user gives an AI system. In AI-visibility work, the set of prompts a brand's buyers actually use, "best CRM for a small business," "alternatives to X," is the equivalent of a keyword list: it defines the questions you want to show up for.
RAG
Retrieval-Augmented Generation. The technique behind most AI search: instead of answering from training alone, the system first retrieves relevant passages from an external source, the live web or an index, and then generates an answer grounded in what it retrieved. ChatGPT search, Perplexity and AI Overviews all use a form of RAG. It is why being in the retrieved sources, not just in the training data, is decisive.
Retrieval vs Training Crawlers
Two different kinds of AI crawler that people constantly confuse. Training crawlers (GPTBot, ClaudeBot, CCBot) collect content to build model training datasets; blocking them has no effect on whether you get cited. Retrieval crawlers (OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-SearchBot) fetch pages for live answers; blocking these removes you from AI results. The practical rule: you can block training and stay fully citable.
SEO
Search Engine Optimization. The original discipline of optimizing a website to rank in a search engine's list of results. AI search does not replace SEO; for Google, AI features largely run on the same index, so being crawlable and indexed remains the entry ticket. SEO is necessary but, increasingly, no longer sufficient on its own.
Share of Voice
In AI search, the proportion of answers in your category that name your brand, measured against competitors. If ten buyer questions in your space produce answers naming brands, and you appear in four of them, your share of voice is a way of expressing that against the field. It is one of the clearest ways to benchmark AI visibility over time.
Zero-Click Search
A search that ends without the user clicking through to any website, because the answer was resolved on the results page itself, by a snippet, an AI Overview, or an instant answer. In the first four months of 2026, 68% of US Google searches ended without a click (SparkToro / Similarweb). Zero-click search is the underlying trend that makes being the answer, rather than a link, so important.
Which acronym should you actually use?
Honestly, it does not matter much. AEO, GEO, LLMO, AI SEO and AIO overlap so heavily that arguing over them is mostly a waste of energy. They describe the same job from slightly different angles: making sure AI systems understand your brand, trust it, and name it. We use "AI visibility" as the umbrella and treat the rest as lenses. If someone insists there is a rigid, important difference between them, they are usually selling one. We lay out the overlap in detail in AEO vs GEO vs SEO.
Frequently asked questions
Want to see how these ideas apply to your brand specifically? Get a free AI Visibility Audit.
Sources
- Google Search Central, "Optimizing your website for generative AI features on Google Search" (chunking, llms.txt, "still SEO") — https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Aggarwal, Pranjal, et al. (2024), "GEO: Generative Engine Optimization," KDD '24, arXiv:2311.09735 — https://arxiv.org/pdf/2311.09735
- SparkToro, "In 2026, Less than One Third of Google Searches Still Send a Click" (68% zero-click) — https://sparktoro.com/blog/in-2026-less-than-one-third-of-google-searches-still-send-a-click/
- BrightEdge, "AI Overviews at the One-Year Mark" (~48% prevalence, 9-industry commercial tracker) — https://www.brightedge.com/resources/weekly-ai-search-insights/ai-overviews-one-year-presence-size-citing
- Google, Search at I/O 2026 (AI Mode 1 billion monthly users) — https://blog.google/products-and-platforms/products/search/search-io-2026/
- OpenAI, "Overview of OpenAI Crawlers" (retrieval vs training) — https://developers.openai.com/api/docs/bots