II for relevant search

Short description

Brave has introduced an AI-powered search technology called Summarizer, which delivers short, concise and accurate answers based solely on web search results. The tool provides up-to-date information in real-time and always provides a link to the source of data, enabling users to assess its credibility. The Summarizer replaces query-based snippets with summarised versions, highlighting the response and aggregating multiple sources of information to create a comprehensive answer. It is based on Brave’s built and managed models, which are optimised for time efficiency and quality. The tool is now available to all Brave search users on desktop and mobile.

II for relevant search

We’re introducing Brave’s new AI-powered search technology, the Summarizer, which delivers short, concise, and meaningful answers to the top of Brave search results upon user request, based solely on web search results. Unlike strictly generative AI models, which often produce unsupported claims, we trained large language models to process multiple sources of information on the web. This allows them to give a much more concise and accurate answer in connected language.

Moreover, the Summarizer always provides a link to where the data came from. This shows the authorship of the information and helps users assess the credibility and degree of trust in the source, which is necessary to combat potential distortions from the authority effect inherent in large language models.

Brave Summarizer is now available to all Brave Search users on desktop and mobile. If you would like to turn off the Summarizer, you can easily do so in the settings.

Using results from the web allows Summarizer to provide up-to-date information in real time. It is important to remember that despite the impressive results of AI models, users should not blindly believe everything that AI does, just as we should not trust everything that is written on the Internet: we should always apply critical thinking to everything information we consume.

In addition to actual summarization, our AI models can replace query-based snippets with summarized versions, highlighting the response where possible. This allows us to provide both a basic summary based on the aggregation of many sources to create a comprehensive answer, and a summary of a single source (eg press articles). We provide both types of summaries together, so users can see both an overview at the top and a snippet with highlighted answers.

It became Note the highlighted answers in the snippets.

“Brave Search is the fastest-growing search engine since Bing: we process 22 million queries per day, providing independent search results based on our proprietary web index. With the AI-powered Summarizer, we’re taking the next step to improve the relevance of search results. Unlike AI chatbots, which can produce fabricated answers, Summarizer provides a well-written summary at the top of search results, aggregating new sources across the web and indicating their provenance for transparency. This open system is now available to all Brave search users for better search navigation,” says Josep Puyol, Director of Brave Search.

Unlike many companies that provide similar functions, we do not rely on third-party technologies and are not limited by issues of scalability. Summarizer is based on Brave’s built and managed models, which are optimized for the most efficient time to generate a result. Today, Brave Search handles daily peaks of 600 queries per second, which we then run through our AI model. So far we are generating resumes for 17% of requests and will increase this number in the near future by scaling our system. We believe that our II-model is the largest such system today, because it gets more traffic in queries per second than others (we apply Summarizer to all queries) and neither Bing nor Google did not open their systems.

In addition to scalability, we invest in the quality of generated resumes. However, since our model is still in the early stages of development, you may occasionally see “chimeras” that mix unrelated snippets into a single result, or offensive or erroneous text, but these will steadily decrease as we progress. we will improve our models and receive feedback from users.

Summarizer was developed entirely by the Brave Search team and is based on our unwavering principles of independence and privacy. We do not use ChatGPT or its backend. Our AI is built on three different large language models that solve different tasks, and the most basic models are built on open source BART or DeBERTa hosted on Hugging Facewhich were heavily retrained on our own search results data.

  • The first model answers questions by trying to get a specific answer from text snippets if possible. Brave already uses large language models to improve the relevance of search results, and this model is an extension of the existing technology that serves our knowledge graph and snippet capabilities. The difference lies in the number and length of analyzed snippets.

  • After the first phrase of receiving answers, received candidates are classified using a set of zero-shot classifiers according to a large number of criteria (hate speech, vulgarity, spam, etc.).

  • Finally, the final set of candidate texts is processed by a summarization/paraphrasing model that attempts to rewrite the input text, removing repetitions and creating a uniform text for readability.

We plan to share more technical details with a particular focus on scalability in the weeks following the full release.

Please note that the Summarizer is currently disabled in the Brave Search Glasses (of our innovative feature that allows users to create their own search engine ranking filters). We will continue to improve our models for quality compatibility with the user’s Glasses, and will share new information about our successes in the near future.


Brave Search’s Summarizer is our next step in improving search relevancy and a response to the recent changes brought about by the release of ChatGPT last December. The news that Microsoft is going to deeply integrate OpenAI models into their Bing search engine has had a big impact on the market: although the system is not yet publicly available, initial feedback speaks of the model and how impressiveand even as about scary.

There’s no doubt that the industry is generating a lot of hype around AI right now, but we at Brave are not yet entirely convinced that big language models will radically transform search. However, when applied correctly, such models can help users engage with search results, which is our approach to Summarizer. At this point, chat-like interfaces are a new technology, and we’re not convinced they’ll work everyone search tasks

Nevertheless, we believe in the potential of large language models and will continue to experiment with their application not only in search, but also in Brave Browser, where we expect truly revolutionary changes thanks to the assistive capabilities of large language models.

Brave Search is available on all browser versions Brave (Desktop, Android, iOS), and is also available in other browsers at search.brave.com.

If you look events in Mykolaiv – https://city-afisha.com/afisha/

Related posts