Google has introduced a new AI tool called Deep Research that allows the Gemini chatbot to conduct a thorough web search on a given topic instead of the user and create a detailed report based on the information found.
Image source: Google
Currently, Deep Research is only available to paid Gemini Advanced subscribers and in English only. To activate it, you must select “Gemini 1.5 Pro with Deep Research” as the model in Gemini.
If you have access, you can ask Gemini to conduct research on a specific topic. After this, the chatbot will create a “search plan” that can be edited or immediately approved. According to Google, Gemini will begin its exploration by “searching for interesting pieces of information” on the Internet, and then perform related searches, a process it repeats several times.
When the study is completed, Gemini will issue a report of its “key findings” with links to the websites where the information was found. The user can ask to expand certain search areas, edit the report, and export the generated AI research to Google Docs. This is reminiscent of the Pages feature offered by AI search engine Perplexity, which generates a custom web page based on a user’s query.
Google unveiled Deep Research as part of its broader announcement of Gemini 2.0, a new model for the era of “agent-based” AI, or systems that can replace the user in routine operations. Deep Research is just one example of Google’s AI agents. Other AI companies are also developing similar solutions.
Along with the Deep Research announcement, Google also announced that Gemini Flash 2.0, a faster version of its next-generation chatbot, has become available to developers.