Web Dandy Web Design Articles
Web Dandy Web Design Articles

Google Overviews — A New Search Addition

Addition Of AI

In a previous article we mentioned the evolution of SEO. This year Google took another step towards improving its search engine with the addition of generative AI Overviews. By harnessing the power of AI, Google aims to deliver more accurate, contextually relevant search results.

This innovative approach promises to revolutionise how websites rank, potentially changing traditional keyword research, link building and on-page search engine optimisation (SEO) forever. As this technology develops, businesses and content creators must adapt to stay visible in an ever-changing search environment, marking the start of a new era in online visibility.

So how does it work? Is it useful, and what does it mean for ranking your website in Google's search results?

Why Add AI?

Google has achieved its search engine dominance by making user satisfaction a top priority and continuously improving their search experience. From its launch, Google ensured it did not repeat the results displayed on a page, which was a fault with many search engines at the time. With Google, you did not waste your time clicking on multiple links that sent you to the same page.

Since those early days, Google has maintained its focus on improving user experience. It was inevitable it would start introducing the capabilities of AI to maintain this edge.

In May of this year, Google unveiled a feature designed to help gather results from multiple websites and present them as AI-generated summaries.

Initially released in the US, Google will roll out this feature in more countries throughout the year. The aim is to make Google the world's first search and answer engine. In effect, to get Google to "do the Googling for you" so you don't have to visit multiple sites from search results to get the answers you are after.

Leveraging A Customised Large Language Model (LLM)

Google's new Overviews feature revolutionises search by leveraging a customised LLM designed to act like a personal AI research assistant. This advanced system uses self-supervised learning techniques and has trained on a vast amount of data to construct a comprehensive understanding of natural language.

By breaking down complex search queries into sub-questions, Overviews can solve each part individually, weighing different perspectives and adjusting its answers based on the time of year and the searcher's location.

Using sophisticated natural language processing to grasp the intent behind a search, it is designed to gather the most relevant information from Google's extensive index of websites into one cohesive summary. This eliminates the need for users to sift through lists of web pages to piece together an answer. Beyond merely merging existing information, it employs advanced reasoning to provide fresh insights and recommendations.

Integrating AI makes Google Overviews a powerful tool for providing more accurate and insightful search results. It aims to improve the fluency and relevance of responses while also reducing the risk of false information.

Does It Work?

Within days of the release of Overviews in the US, it became clear that there were issues with the answers being given by Overviews.

Possibly one of the most infamous Overview responses occurred when someone asked how to prevent cheese coming loose from the top of a pizza. A humorous post on Reddit was used as the source of Google Overviews. The AI suggested mixing non-toxic glue into the cheese to avoid the problem. Unfortunately, the answer-generating step did not correctly interpret the irreverence of the post.

Misunderstanding humour also explains the use of a satirical website, to answer a query about diet, to recommend eating at least one small rock per day as part of a balanced diet.

Other examples of poor answers stated that Barack Obama, a Christian, was the first Muslim President of the United States, and that US President Andrew Johnson gained university degrees between 1947 and 2012. He died in 1875.

Predicting The Next Word

When an LLM produces text by only predicting the next most likely word, it can produce a sentence containing untrue information referred to as hallucinations.

So what has been going wrong?

To produce a suitable answer, Overviews has to complete two steps. Step one is to retrieve information correctly, followed by step two, to generate a correct answer. If one or both steps fail, it can cause a poor answer.

It is known that large language models can encounter a problem when they come across conflicting information e.g. if it finds two editions of a handbook or document, it cannot decide which version to use and may combine both, creating a potentially incorrect answer. Google says the software may also produce some incorrect answers if there is a lack of high-quality information on the web or if a query closely matches satirical sites or humorous posts.

Hallucinations are going to be a possibility so long as probability is used to generate text word by word. Especially if the software struggles to separate serious facts from fiction, satirical or otherwise. The errors highlight the need for greater scrutiny and thorough testing. So how has Google responded to the issues?

Liz Reid, Head of Google Search posted an article with an update to Google's AI use in Overviews discussing some of the queries that had been returned saying "some odd, inaccurate or unhelpful AI Overviews certainly did show up. And while these were generally for queries that people don't commonly do, it highlighted some specific areas that we needed to improve."

As a consequence Google has added restrictions for queries where AI Overviews have not proven to be especially helpful and they have also added additional refinements for queries related to health. Google will no longer show AI Overviews for dangerous topics or for queries that suggest a vulnerable situation.

The Implications For Ranking

Despite a shaky start, Google's addition of Overviews into its search engine is likely to lead to a significant shift in how websites will be ranked in the future. Google aims to enhance the AI's ability to summarise information accurately and adapt its replies to individual users, to move from merely answering queries to facilitating ongoing, in-depth conversations. They want users to explore topics through a free-flowing dialogue with the AI, similar to speaking with an expert.

As the use of AI in search develops, SEO strategies will have to develop too. The focus will increasingly be on high-quality, informative content, with an emphasis on relevance and user intent over traditional keyword density. SEO will have to depend on understanding AI search algorithms and optimising for content quality and relevance. It will have to prioritise question-based keywords and detailed content rather than merely focusing on keywords and backlinks.

Google may have taken the first steps in disrupting search engine optimisation with its new AI-based feature. As AI alters the future of search, businesses and content creators will have to adapt to these changes to maintain visibility and competitiveness in the online landscape.