How AI is impacting search engines 

By far the biggest news in web technology in recent times is AI, and unsurprisingly Google is at the forefront of developments. The search engine is undergoing significant changes driven by advancements in generative AI, which may prove a transformative period for the platform, or possibly herald its decline – OpenAI has just announced its SearchGPT product to challenge the existing order.

The most notable update to Google is the introduction of AI Overviews (previously called Search Generative Experience or SGE), designed to streamline the user experience by summarising complex queries directly in the search results page. These overviews use Google’s Gemini model, which combines advanced capabilities such as multi-step reasoning and multimodality with traditional search functionalities. In short, that means it can interpret data from images and videos, including the ability to search with a video rather than a text query (in the same way you can already search using an image), whilst also understanding complex queries such as “find the best Indian restaurants in Birmingham and show me details of any offers they have, with walking time from the Bullring.”

Clearly, AI Overviews could drive even higher reliance by users on Google’s ecosystem by centralising information within their platform, as users no longer have to leave a Google site or app to get the information they seek. This poses challenges for businesses and regulators, with Google (and other tech companies) already facing big legal cases related to their near-monopoly status. There is also the question of copyright that is such a thorny issue for all AI platforms, with the majority having crawled the web to get the data that feeds their models without any permission from the copyright owners to use their data.

Google faces challenges in all of this too: Firstly, generative AI is famous for its hallucinations, where completely incorrect statements are given as fact. This is inherent to how the models work, so Google will have to be very careful policing what AI Overviews generate (it has already told people to glue cheese to pizza and to eat rocks).

Secondly, Google has to find a way to monetise these new features, as they have the potential to take attention away from ads both in terms of placement (at the top of results) and in terms of clicks (as users get the answer they need without having to click on anything else). Google is now testing ads in AI Overviews. Interestingly, the number of search results featuring AI Overviews appears to have dropped significantly since they were first rolled out (from 11% in June to 7% now).

In addition to AI Overviews, Google is integrating new planning capabilities within its search engine. In the US, users can now generate meal plans, travel itineraries, and other detailed plans directly from search queries. This feature allows for easy customisation and exportation of plans to Google Docs or Gmail, enhancing the utility of Google Search as a planning assistant.

Google is also introducing AI-organised search results pages, which categorise information under AI-generated headlines to make exploration easier. Initially focused on dining and recipes, this capability will extend to other areas such as movies, music, books, and shopping.

Meanwhile, Microsoft has announced its own generative search experience in answer to Google’s AI Overviews. As Google does, Microsoft also claims that the experience “maintains the number of clicks to websites”. Bing already had its Copilot feature, which would provide answers as an additional source alongside the “traditional” search results, but this new experience swaps things round – the AI-generated information is now where the main results appear, whilst traditional search results of links to web pages are relegated to the right-hand side.

With all these AI-driven features coming to search engines, you might wonder how they treat AI-generated content on other websites. In short, they’re agnostic – in their guidance about AI-generated content, Google says “Appropriate use of AI or automation is not against our guidelines. This means that it is not used to generate content primarily to manipulate search rankings, which is against our spam policies.”

Essentially, if you use AI to produce high-quality content that is of value to your visitors, it’s no different to if a human had written it. However, if you simply churn out content for the sake of it with no added value, personal insight or opinion, you’re just regurgitating information already available and that might not fare so well with search engine algorithms.