Is AI a Website? Models, Servers, and Web Pages Compared

Why a chatbot page is not the AI behind it: definitions from W3C and NIST, how a request reaches a model, and how web pages feed AI training data.

Guide: AI BasicsLast verified: October 20266 min read

No. A website is a collection of linked web pages served from one network location, while artificial intelligence is a kind of system that generates predictions, recommendations, or decisions from inputs. The confusion comes from the fact that many people meet a chatbot through a web page, but the page is only the front door. The AI itself is a trained model running on servers, and the same model can also be reached through an app or a programming interface.

Key takeaways

  • The W3C glossary, citing a 1999 terminology sheet, defines a web site as a collection of interlinked web pages, including a host page, at the same network location.
  • NIST's AI Risk Management Framework 1.0 (2023) treats an AI system as a machine-based, engineered system whose outputs (predictions, recommendations, or decisions) serve a set of objectives.
  • A chatbot site has at least three parts: a web page in the browser, a server that answers requests, and the model that produces the text.
  • Websites and AI do relate in the other direction. Text gathered from public web pages is a common source of training data, and site owners can signal crawlers with robots.txt.
  • The robots.txt standard, RFC 9309 (September 2022), states that its rules are not a form of access authorization.

What each term means

The W3C glossary defines a web site as "a collection of interlinked Web pages, including a host page, residing at the same network location." It takes the wording from a 1999 W3C terminology sheet. The definition is about structure and location: pages, links, and a place on the network.

The definition of an AI system comes from a different tradition. NIST's AI Risk Management Framework 1.0 (2023) frames an AI system by function: a machine-based, engineered system that, given objectives, yields predictions, recommendations, or decisions that can shape real or virtual environments. NIST notes the wording is adapted from the OECD's 2019 recommendation and from ISO/IEC 22989:2022. This definition is about what the system does. It says nothing about where the system lives or how people reach it.

The two definitions answer different questions, so one cannot be a case of the other. A site can contain AI, and AI can be delivered through a site, but neither contains the other by definition.

What happens when a chatbot page loads

Under the Hypertext Transfer Protocol, described in RFC 9110 (June 2022), a client is a program that connects to a server to send requests, and a server is a program that accepts those connections to service the requests. The browser is the client. The target of each request is called a resource, such as a page, an image, or a block of data.

On a chatbot site, the browser first receives ordinary web pages: layout, buttons, and a text box. When a message is typed, the page sends it to a server. The server may check an account, apply usage limits, and pass the text to the model. The model returns text, and the server sends it back to the page for display. Only the first and last of these steps belong to the website. The middle step is where the AI system sits.

This split explains why one model can power several products. The same underlying system might be offered through a browser page, a phone app, a plug-in inside another program, or a programming interface that developers call from their own code. In each case the model is the same kind of thing, and the delivery channel changes.

Three layers that get mixed up

The table separates the layers. The examples are generic and illustrative, not tied to any product.

LayerWhat it isIllustrative exampleIs it a website?
ModelA trained set of numbers (parameters) plus the code that applies themA file of parameters loaded on a serverNo
ServiceServer software that receives requests, runs the model, and returns outputA programming interface that accepts text and returns textNo
InterfaceWhat a person uses to reach the serviceA chat page in a browser, or a phone appOnly when delivered as linked web pages

The other direction: websites as AI training data

The link between websites and AI is stronger on the data side. In the 2019 paper that introduced the T5 model, Colin Raffel and colleagues describe building a training set called the Colossal Clean Crawled Corpus from Common Crawl, a public web archive that produces around 20 terabytes of scraped text each month. After filtering, the set was about 750 gigabytes. The paper notes that most of the raw text was not usable natural language, and the authors removed lines without terminal punctuation, pages with boilerplate, and duplicate passages. Public web pages are therefore raw material for some AI systems, which is a different relationship from being one.

Site owners have a standard way to state preferences to automated crawlers: the Robots Exclusion Protocol, specified as RFC 9309 in September 2022 by M. Koster, G. Illyes, H. Zeller, and L. Sassman. It asks crawlers to honor rules published in a robots.txt file. The document is explicit that these rules are not a form of access authorization, so the file expresses a request and does not enforce it.

A quick test for any "AI website"

  • Open the page and look for the model's name. A site that only wraps another service should say what runs behind it.
  • Ask whether the same service is offered outside the browser, such as in an app or an interface for developers.
  • Check where the data is sent. Text typed into a page travels to a server operated by someone.
  • Read the terms and privacy notice for retention and training uses of submitted text.
  • Treat a site's claim to be "an AI" as marketing language until the underlying system is identified.

FAQ

Is ChatGPT a website?

ChatGPT has a website, but the product is the AI service behind it. The page is one way to reach that service. The distinction applies to any chatbot offered through a browser.

Can a website use AI without being AI?

Yes. A site can include features such as recommendations or search ranking that rely on trained models while remaining a collection of web pages. The AI is a component of the site's back end.

Can AI work without the internet?

Some models can run on a local computer, because a model is a file of parameters plus code. Large models need substantial hardware, and many products run on remote servers for that reason.

Does robots.txt stop AI from reading a site?

It signals a preference. RFC 9309 says the rules are not access authorization, so a crawler that ignores the file is not technically blocked by it.

Last verified: October 2026

Sources

  • Web site, W3C Glossary and Dictionary, citing the Web Characterization Terminology and Definitions Sheet, 1999
  • RFC 9309: Robots Exclusion Protocol, Internet Engineering Task Force, 2022
  • Artificial Intelligence Risk Management Framework 1.0 (NIST AI 100-1), National Institute of Standards and Technology, 2023
  • RFC 9110: HTTP Semantics, Internet Engineering Task Force, 2022
  • Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer, Raffel et al., arXiv, 2019

New to artificial intelligence?

Start with the basics, then follow the guides to tools, rules, and effects on society.

Start with AI Basics