As you can guess from the name, the tool is designed to do a thorough search on the web for information related to your query, then present a detailed report to your specifications.
According to OpenAI, Deep Research “leverages 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”.
Right now, you need to be a paying ChatGPT user to access Deep Research, so you’ll have to give OpenAI at least $20 per month to make use of it. To date, there’s been no word on if or when the feature will make its way to users on the free ChatGPT tier.
Click Deep research to enable it, then compose your prompt. This can be as simple or as complex as you like, but as with normal ChatGPT conversations, more detailed prompts are generally going to give you better results. You can ask about whatever you want, whether a complex science question, the history of a music genre, a trend in the smartphone industry, or the context behind a work of literature.
Hit Enter to submit your prompt, and the conversation begins. The initial responses you get are going to depend on the prompt you’ve used, but ChatGPT might ask you for clarification on certain points in your prompt, or for some guidance as to the different directions you want the research to go in.
When ChatGPT has got enough information to go on, the research starts—and you’ll see a dialog on screen telling you that. While the AI bot is working, you can switch to different websites and apps, or close down ChatGPT if you need to, because the research carries on in the background.
If you want to see ChatGPT’s workings, click on the progress bar on screen: This opens a new panel that tells you about the actions ChatGPT is taking and the reason for those actions. You can also see a list of web searches that ChatGPT is working with. How long you’ll need to wait for the final report depends on the question you asked, but many projects take around 10 minutes.
When ChatGPT has finished its research project, it’s included as a conversation in the current chat—it’s not saved in a separate document, though you can use the copy button after the report (the two squares) to send it to the clipboard and move it somewhere else. You’re also able to use the thumbs up and thumbs down buttons to give feedback on the finished report.
ChatGPT should break up the research for you using subheadings, images, and tables, though this will vary depending on what you’re asking about. You’ll also see web links embedded in the text; you can follow any of these links to see where the bot got its information and to double-check facts (which you should always do when using AI programs).
You can then carry on the chat as you would with any other conversation. You might want to ask for more information based on what you’ve read, for example, or for clarification on some of the points in the report. You can even start a new research report by clicking on the Deep research button again.
Beneath your report you’ll see another icon that looks like a pencil with a plus symbol next to it. Click on this and you get switched to the canvas mode inside ChatGPT, which is more suitable for working in partnership with the AI on the text—the canvas interface makes it easier to add your own edits or to insert more AI research into what’s already there.
Compared with Deep Research in Google Gemini, you get pretty similar results in ChatGPT. The Google interface gives you more control over the specifics of the research before the AI gets started and more prominently shows the sources being used, but it doesn’t show its ‘thought’ processes as clearly as ChatGPT does. Ultimately they both can be useful as research tools—and both are reliant on the quality of the web sources they find for the usefulness of the final reports.
Leave A Comment