After months of hiatus, Microsoft has relaunched its Bing Deep Search

Deep Search feature utilizes Microsoft’s powerful GPT-4 language model, which will provide more insightful and comprehensive search results to users compared to what made up information GPT-3.5 reveals. 

Deep Search acts as an extra layer on top of Bing’s existing search engine. It uses the vast search index and ranking system while incorporating GPT-4.

• GPT-4 analyzes search queries and refines them into a more comprehensive description of the user’s intent, which surfaces relevant information that might not be readily apparent in traditional searches.
• By employing various “querying techniques,” Deep Search explores different facets of the expanded query, potentially leading to a broader and more informative set of results.

We got this info from X.

It looks like they have also addressed previous caching issues, and the relaunch has enhanced overall stability, which is resulting in improved performance, as per user feedback. 

It also provides a more transparent experience for users, so Copilot informs users upfront about the 30-second processing time for GPT-4 powered results, which goes well with the promise they made earlier.

Deep Search’s initial test launch generated mixed feedback, primarily due to performance shortcomings. Microsoft later on admitted its unanticipated popularity.