Artificial Intelligence (AI)

Artificial intelligence is the field of building computer systems that carry out tasks which would otherwise call for human judgement: recognizing, classifying, predicting, translating, and producing language. In this glossary the term nearly always means the large language models behind assistants such as ChatGPT, Claude, Gemini and Perplexity.

The layers inside the term

AI is the outermost word. Inside it sits machine learning, where a system learns a behavior from examples instead of being given rules. Inside that sit neural networks, the family of models that learn by adjusting millions or billions of numeric weights. Large language models are neural networks trained on text to predict what comes next, and generative AI is what they are called when the output is new text, images or code rather than a label. A spam filter is AI. So is a chess engine written as rules in 1990. The word alone tells you very little about what the system does.

What the word means when a business owner uses it

The question this site is asked is whether AI recommends a business. That question is about one specific kind of system: an assistant that takes a question in plain language, may search the web and fetch pages, and writes an answer that names a few businesses. It is the only part of AI this site measures. Ad bidding, route planning and the recommendation engine behind a shopping site are also AI, and none of them is what the question is about.

The assistant’s answer is built from two things: what the model absorbed in training, which is fixed at a cut-off date, and what it fetched a moment before answering. For a small business the second matters more, because it is the part the business controls. What its site serves to a crawler is what the assistant reads.

A common question

Which part of AI actually decides whether a business gets named in an answer?

One narrow part of it: an assistant that takes a question in plain language, may search the web and fetch pages, then writes an answer naming a few businesses. Ad bidding, route planning and the recommendation engine behind a shopping site are AI too, and none of them is what the question is about. An assistant's answer comes from two places, what the model absorbed in training, fixed at a cut-off date, and what it fetched moments before replying. The second matters more to a small business, because it is the part the business controls.

Sources

Definitions here are stated from mechanism and carry no figures. The measurement this site runs against assistants is described on the methodology page.