ChatGPT’s New Deep Research Feature Unveiled
ChatGPT continues to evolve, adding new capabilities to its repertoire. The latest feature, “Deep Research,” allows users to conduct thorough web searches and generate detailed reports. This tool is similar to a feature offered by Google Gemini, enhancing ChatGPT’s ability to provide comprehensive information.
According to OpenAI, Deep Research is designed to “leverage reasoning to search, interpret, and analyze massive amounts of text, images, and PDFs on the internet, pivoting as needed in reaction to information it encounters.”
Currently, Deep Research is available exclusively to paying ChatGPT users, requiring a subscription of at least $20 per month. There has been no information released about when, or if, this feature will become available to users on the free tier.
Starting a Research Project
The Deep Research tool is accessible via the ChatGPT web app, desktop, and mobile interfaces. In the web version, users can initiate the process by selecting a model that supports this feature from the dropdown list at the top left of the interface. Once selected, you’ll find a “Deep research” button in the main prompt box.
By clicking the Deep research button, you enable the feature, allowing you to compose your prompt. Prompts can range from simple to complex, but generally speaking, more detailed prompts result in better, more substantial results. You can ask almost anything here – from complex scientific questions, historical events, or even popular culture.
Once submitted, ChatGPT will begin its research and may ask for clarification or guidance to better understand the scope and focus of the research. As the AI works, a dialog box will appear informing you of its progress. This allows you to continue working on other tasks without interrupting the research process.
If you wish to monitor ChatGPT’s activities during the research process, clicking the progress bar on the screen will open a panel detailing the actions the AI is performing. This window also presents a list of the web searches being used.
The time it takes to complete the research varies, but the report is usually generated in around 10 minutes.
Reading Your Research Report
When ChatGPT finishes, the research results are provided within your current chat, not as a separate document. However, it allows you to copy the final report and paste it elsewhere using the two square copy icon. Users can also provide feedback on the report using the thumbs up and thumbs down options.
ChatGPT typically organizes the research using subheadings, the inclusion of images, and tables may also appear in the report, depending on the search query. Web links are embedded within the text; these links enable users to verify facts by visiting the source data.
After reading, you can continue the conversation just like any other ChatGPT interaction to gather more details and information, or get clarification on any points in the report. You can then launch a new research project using the Deep Research button.
Users can click on the pencil icon to the left of the report to enter canvas mode. This interface is designed to let you add your edits to the text and insert more AI research.
