Microsoft is integrating a new AI function into its Bing search engine that is intended to save users from having to make precise queries. The feature, based on the GPT-4 language model, is called “Deep Search” and is designed to expand the user query, allowing the search engine to provide more relevant results.
As an example of how deep search works, Microsoft showed how Bing turns a vague query into “How do point systems work in Japan?” More detailed instructions about the search engine can be found here. The result of a short query was the following:
“Explain how different loyalty card programs work in Japan, including their respective benefits, requirements and limitations. Add examples of popular loyalty cards from different categories such as shops, supermarkets and restaurants. Show a comparison of the pros and cons of using loyalty cards compared to other payment methods in Japan, including current promotions and incentives. Highlight the most popular services and participating sellers..
Additionally, Bing can expand this query to include descriptions of immigration policies and public transportation fares, as GPT-4 attempts to do “Find all possible targets” Inquiry. Whatever topics the search engine associates with the user’s query, they will all be displayed in the panel on the search results page. In the future, the user can choose which of the suggested topics best suits their request. “Deep Search” then processes the data further and organizes the results “depending on how well the results match the detailed description”.
According to Microsoft, processing a request with the new feature can take up to 30 seconds. If the user doesn’t want to wait that long, they can reject the innovation and continue to interact with the regular Bing search.