How Does Bing AI Search Make Use of Website Content?

The Microsoft Bing Blogs have published new details about how the new AI-integrated search works, including information about a technology called Bing Orchestrator, which acts as a coordinator between the Bing website index and the "next generation GPT" underlying the new version of Bing.

This has two sides: the front end and the back end.

Bing is blending links to website sources within a relevant contextual output on the front end. That's good news for publishers because it means better search referrals without the ambiguity of the traditional ten blue links.

The back end includes the Bing Orchestrator, which uses web data from the Bing index to ground the GPT part, improving relevance and benefiting publishers.

Prometheus Model from Microsoft

Microsoft previously described the Prometheus AI Model, which is a collection of methods for interacting with the OpenAI model.

The most recent blog post contains additional information about the new Bing search with AI.

The first thing that stands out is that OpenAI shared its next-generation GPT with Microsoft in the summer of 2022.

This next-generation model has "creative reasoning capabilities" that outperform GPT-3.5, the technology that underpins ChatGPT.

This model inspired Microsoft researchers to develop the Prometheus AI model to interact with it.

Prometheus is a hybrid of the Bing search index and what Microsoft refers to as the Next Generation GPT, which is managed by the Bing Orchestrator.

Bing Orchestrator's Grounding

The Bing Orchestrator sits between the search and the GPT components.

Bing Orchestrator takes the search query and coordinates the process of retrieving an answer from the Bing index and GPT.

The breakthrough is how Bing search and answers provide "grounding" for GPT.

Grounding is a technique that enables the AI to use that additional data to provide more context to the answer, specifically context related to what is happening right now, in a geographic region, or in other contexts.

In general, grounding is a method of giving words context. Language models understand words in terms of their relationships to other words, such as this is related to that.

Grounding, in the sense that Bing employs it, adds context.

According to Bing's announcement:

"…it provides relevant and fresh information to the model, enabling it to answer recent questions and reducing inaccuracies - this method is called grounding.

In other words, the model reasons over Bing data and is thus grounded in Bing data via the Bing Orchestrator.

Prometheus can also integrate citations into sentences in the Chat answer, allowing users to easily click to access those sources and verify the information, thanks to the Bing grounding technique."

That last part about adding citations to website sources is critical because it transforms the chat answer into a Wikipedia-style answer with links to more information.

Those are contextually relevant links, which is an unusual but sensible way to combine AI and search.

Bing Orchestrator can also generate iterative queries.

This is what the new blog post says about the AI-enhanced Bing:

"Through a component called Bing Orchestrator, Prometheus leverages the power of Bing and GPT to generate a set of internal queries iteratively, with the goal of providing an accurate and rich answer for the user query within the given conversation context."

Bing does not go into specifics about iterative queries. The term iterative refers to a process of repetition.

It presumably refers to repeating a query in order to generate a larger set of answers and context for the GPT.

Inside Consider the AI-Enhanced Bing

The blog post contributes to our growing understanding of what the future of AI-enhanced search will look like.

Many people are concerned because when they think of AI-based search, they envision a machine that provides answers but no links to websites.

However, that type of model is more akin to what is known as Artificial General Intelligence (AGI).

An AGI is a machine that can learn and know everything like a human.

AI is still in its infancy.

Bing's blog post demonstrates how AI improves search in the same way that featured snippets and knowledge graph data improve search.

AI is already on the hunt. Google, for example, employs AI to detect spam (SpamBrain is an AI-based spam detector).

Another Bing article frames AI-enhanced search as a co-pilot (Reinventing search with a new AI-powered Microsoft Bing and Edge, your copilot for the web).

This includes both the backend (internal) and front-end (user-facing) aspects of search.

The article Search as a Co-pilot explains:

"Model Microsoft Prometheus. We have created a proprietary method of working with the OpenAI model that allows us to maximize its power.

This set of abilities and techniques is known as the Prometheus model. This combination provides more relevant, timely, and targeted results while also improving safety.

...Using AI to improve the core search algorithm. We've also applied the AI model to our core Bing search engine, resulting in the largest increase in relevance in two decades. Even basic search queries are more accurate and relevant with this AI model."

Returning to how the integration of GPT into search is more of an enhancement to the search experience, Microsoft sees GPT integration as part of the ongoing evolution of search beyond the classic ten blue links, which cannot provide the rich answers that complex search queries require.

According to the Microsoft search co-pilot article:

"An entirely new user experience. We're reimagining how you interact with search, browser, and chat by bringing them all together in one place. This will enable a completely new way of interacting with the internet."

Hocalwire CMS includes fantastic automation features that help traffic from various sources focus on your website. You may drastically increase your traffic with the aid of Google Analytics and potential choices with Hocalwire. To learn more about Hocalwire CMS's limitless potential, schedule a demo.