Google boosts AI search with improved AI Overviews, new AI Mode

Google is expanding generative AI’s role in search with an improved AI Overviews, which can handle coding, advanced maths, and multimodal queries. AI Overviews will now appear for more queries and will be available to teenage users. 

Google is also testing a new feature called AI Mode, which can address complex, multi-part queries and respond to follow-up questions. It will be available on an opt-in basis to Google One Premium AI subscribers for feedback. 

“This new Search mode expands what AI Overviews can do with more advanced reasoning, thinking and multimodal capabilities so you can get help with even your toughest questions,” said Robby Stein, VP of Product at Google Search, in a blog post Wednesday.

To bake the new AI mode into Search, Google is combining the generative AI capabilities of Gemini 2.0 with its well-established information systems that includes search index, knowledge graph (database of people, places, things), real world data, and shopping data of billions of products. 

The AI Mode will use a technique called ‘query fan-out,’ which breaks down a search query into subtopics and simultaneously issues multiple searches across those subtopics and data sources, instead of processing it as a single query.

“This approach helps you access more breadth and depth of information than a traditional search on Google,” said Stein. 

Google is keeping the AI Mode close, keeping in mind the backlash it faced after the initial release of AI Overviews last year.  During the first public release, AI Overviews responded to search queries with erroneous suggestions such as eating rocks for their nutrient value and applying glue on pizza. It raised questions about the potential risk of AI and how it can be magnified by search given its wider reach. 

Keeping these concerns in mind, Google plans to revert to showing web search results and web links for queries where it’s not sure AI Mode can be helpful or provide a quality response. Google also wants to make AI Mode more visual in its responses and use more images and videos instead of just text. 

While Google still dominates the search business with 90.14% market share, as per Statcounter, the arrival of generative AI has changed how people seek information online. 

Rise of search-focused AI apps such as Perplexity is drawing away many users from traditional search. 

During the recently concluded US presidential elections, Perplexity successfully provided real-time election coverage and analysis using data from Associated Press and Democracy Works (a non-profit). 

Google’s biggest rival Microsoft was one of the first to tap generative AI to improve the search experience and started to offer summaries and curated responses to search queries, instead of just showing weblinks. OpenAI’s ChatGPT now also offers built-in search capabilities that allows users to find real time information on any topic.

Adding AI to search has simplified the search experience for users. Instead of wasting time going through multiple web links to find answers, users can get AI-curated responses drawn from multiple web links and articles within seconds. 

Though useful, AI-driven search has also raised copyright infringement lawsuits from newspapers and publishers, which claim that their copyrighted articles are being used without permission or remuneration by these AI platforms to respond to user queries. 

This has forced the AI firms to sign licensing deals with several leading news publishers that include payments as well as showing links to the original article. OpenAI has signed five-year partnership estimated to be worth $250 million with News Corp that publishes Wall Street Journal, Dow Jones, The Sun, and The Times. 

Other news publishers that have signed deals with OpenAI include Conde Nast, Axel Springer, Associated Press, Financial Times. 

Image credit: Google