Microsoft is gearing up to introduce a new feature to Bing called Deep Search, designed to simplify the process of creating AI prompts. This GPT-4-powered capability takes users’ Bing queries and expands on them, enabling the search engine to retrieve comprehensive information related to the question from the web.
Illustrating its functionality, Microsoft demonstrated how a generic query like “How do points systems work in Japan” can be transformed into a detailed prompt. The refined prompt asks Bing to explain various loyalty card programs in Japan, outlining benefits, requirements, and limitations. It also includes examples from different categories such as convenience stores, supermarkets, and restaurants. The prompt further seeks a comparison of using loyalty cards versus other payment methods, highlighting current rewards and benefits, and spotlighting popular services and participating merchants.
Bing, leveraging GPT-4‘s capabilities, may broaden the search scope to encompass additional topics like immigration policies and public transportation fares, aiming to explore all possible intents behind a user’s query. The results of the Deep Search will be presented in a panel on the search results page, allowing users to choose the most relevant information. The search results are curated and ranked “according to how well they match the comprehensive description.” Microsoft acknowledges that conducting a Deep Search may take up to 30 seconds, but users always have the option to revert to Bing’s standard search feature if they prefer quicker results.
Although Deep Search is currently in an experimental phase and limited to a small user group, Microsoft introduced it in conjunction with Copilot’s support for OpenAI’s GPT-Turbo model. This innovation aims to streamline the process of formulating AI prompts, addressing the challenge of crafting queries and enhancing productivity for users.
