Table of Contents
Google hasn’t been shy about rolling out new AI tools and features in recent months, from Gemini to Gmail AI-hosted podcasts. One of the latest innovations unveiled is Google Deep Research, which essentially unleashes Google’s Gemini AI onto the internet, with the mission of deeply investigating a topic of your choice.
Imagine doing something that would normally require a lot of Googling: finding the best phone to upgrade to, trying to understand how a self-driving car works, or mapping Scotland’s history in the 17th century. Deep Research can handle any challenge like this.
It’s what’s known as an ‘agentic feature’: a trending term in AI that basically means giving these bots more choice and control over what they do. The idea is that there’s less hand-holding to do and the AI to work more independently, which (in theory) should mean less work for you to do.
“Under your supervision, Deep Research does the hard work for you,” Google explains. “After you enter your question, a multi-step research plan is created for you to review or approve. Once you agree, it will begin in-depth analyzing relevant information on the Internet on your behalf.”
One caveat: you must be signed up for the $20 per month Gemini Advanced subscription to try this out now (which also gives you 2TB of Google One storage). Go to the Google One plans page, and you may get access to a free trial of Gemini Advanced, if you haven’t signed up previously.
Get started with Google Deep Research
There are three main phases in the Google Deep Research workflow: To go to the first, go to Twins on the Internetchoose the 1.5 Pro with in-depth research option from the drop-down menu in the top left corner, then tell the AI what you’d like to learn more about (you should see some suggestions appear if you run out of inspiration).
What you look for is entirely up to you, but this is a tool for extensive research: it should go beyond something that would get you a standard AI chatbot answer (such as “what is a good board game for 5 year olds” or “how to you get wine stains from a fabric sofa”). You can think about something you need help with for work or as an extra income, or just something you are interested in as a hobby.
Enter your prompt in the box (it doesn’t have to be long, but the more detailed the better, if possible) and the first phase of Google Deep Research will begin. The AI will think about the assignment you’ve given, think about how best to approach the topic, and then give you an overview of how the AI will approach the challenge.
This research plan is presented to you and you can make changes by clicking on it Edit plan: You’ll then be put back into a chat with the AI, so you can tell it to focus more on certain areas, or add additional research areas it wants to cover. Each time you make a suggestion, you will receive a revised plan.
When you see a research plan on the screen that you are happy with, click Start research to proceed to the next phase. Gemini gets to work and shows you on the screen which websites it is viewing and what information it is analyzing. You can sit and watch, or you can move on to something else. You can even close the Gemini app completely if you need to, and come back later to get your answers.
Answers and reports from Google Deep Research
You won’t have to wait too long for your answers (many questions can be researched in just a few minutes) and you will then be presented with a detailed report, which should cover everything in your original question. If you asked specific questions about a topic, they should be answered somewhere specific.
This is the third phase of Google Deep Research, where you work through the report. However, it doesn’t have to be the end of the investigation: you can ask follow-up questions about anything in the report, just as you would in a normal conversation with Gemini. If you continue chatting, the report itself will be minimized, but you can reopen it if necessary.
At the end of the responses, and at the end of the report itself, you will be given references to the Internet: these can be useful for checking the operation of Gemini and searching for further information yourself. Even in Deep Research mode, the message “Gemini can make mistakes” is displayed, and it’s important not to rely too heavily on AI reporting until you’re sure everything is correct.
Another useful feature of Google Deep Research: you can convert your report into a Google Docs document by clicking the button Open in Documents button in the top right corner. This will give you a fairly simply formatted document that you can edit or save as usual. If there are tables in your report, they will be transferred to Google Docs and you will still get all your web references at the bottom.
From testing Google Gemini Research on a few technical topics, it seems to do a pretty good job at gathering information and figuring out what’s important and what’s not. Having access to the web resources the AI used really helps dig deeper and check facts. Your mileage may obviously vary depending on the topic and information available, but the tool certainly has the potential to save you a significant amount of research time.
Win the holidays with PopSci’s gift guides