Entity-based AI search or (Open AI, Google generative search, and voice search), systems use the entity (relationship model) data to find fact and documents.
One can say a human currated list is the best, but it is not scalable. It takes to many hours for humans to go through 1000s of pages, become and expert on the subject and then direct people to the correct document based on their expertise.
Necessity is the mother of invention
India had a problem, a need, one of their branches of Government had so many pages and a traditional keyword based search did not provide satifactory results for their clients. The labor to help these people navigate the bureaucracy was considerable.
The size of the document pool was small enough however that AI and Natural Language Processing could be a solution. They worked with the open source AI community to build a solution; It worked and the rest is history.
Smart people have the superpower to make simple complicated
AI is scraping the data and creating an entity-relationship model. It looks at a sentence and finds the subject, finds the verb, finds the modifiers, and finds the attributation. The entire language model is beyond the scope of this post.
AI reads, "Honey is sweet," and created the relationship, "Honey sameas sweet."
It reads "Oaked Chardonnay is a type of Chardonnay" and created the relationship, "Chardonnay hasType Oaked Chardonnay."
To answer the question it looked through its data for the types of Chardonnay and found a type that is the same as sweet.
Even though AI did not read, "Oaked Chardonnay is generally sweeter than unoaked Chardonnay," it can artificially look intelligent and provide the answer.
The entity-relationship model, a by-product of AI, can be used independently of the computer that created the model.
Entity-based search explained
Say you ask a question, "Which Chardonnay is more sweet?"
Within the entity data for Chardonnay there are two major types of Chardonnays that effect its flavor profile.
Entity: Chardonnay
Description: Chardonnay grapes can be made into a range of wines, from bone-dry to sweet dessert wine. Even if a Chardonnay is made in a dry style, there are several factors that can make it seem sweet.
TypeOf: Oaked Chardonnay
Description: Aged in oak barrels, this wine has a rich texture, full body, and sweet aroma with notes of butterscotch and vanilla. The palate offers a buttery flavor with notes of honey, hazelnut, and caramel.
TypeOf: Unoaked Chardonnay
Description: Aged in stainless steel tanks, this wine has a light body, bright color, and crisp minerality. Its nose has citrus aroma notes with hints of lime, apple, and peach,
AI can answer, "Oaked Chardonnay is generally sweeter than unoaked Chardonnay," and provide references.
The concern from webmasters, about whether or not SEO is Dead, is more about human behaivor than AI based search. If people only want the an answer prehaps they are in a discussion about a fact, maybe they wont visit a site? The answer is all they wanted. Fair enough but if that is the case would they have done anything after getting the answer, if they had visited the site?
People who want more information from the source or the horses mouth will still visit the site and those who want to continue to get information will sign up for a newsletter ... those who want to purchase a product will go to the source to purchase the product. The question for SEO is which source is AI search going to reference. If it fails to reference a source at all, people wanting to purchase a sweet chardonnay will go to a search engine.
Schema and AI Based Search
Schema was created to make content more readable by machines. The information to build the Natural Language data base is made unambigous in schema. The about property (for what the page is about) references a thing that can be any entity.
Using schema is a SEO best practice as search engines are able to use the natural language by product of AI.
Entity-based search is a disruptive technology
Searching based on keywords, "Chardonnay and Sweet" is likely to surface pages where sweet desserts are paired with Chardonnay. Clearly entity-based algorithms are better for this question to both answer the quesiton in a summary format and provide links to relevant pages. The words "which, is, more," are not relevant keywords and require natural language processing, (which is the part of AI that builds the entity database), to understand the intent.
Maybe trying rephasing the query to, "sweet, flavor, Chardonnay, -dessert," would find the answer in a only keyword based search?
A search algorithms is based off of mathmatics and ranking factors must be quantifiable or countable ... factors that are only true or false can be used for quality indicators to determine if a site will be listed or buried.
