Google expands availability of Deep Research, an AI tool that wades deep into any topic – News

Google LLC said today it’s making its experimental artificial intelligence research tool, called Deep Research, available to Gemini Advanced users via the Google Workspace platform.

A relatively new feature that was announced last year, it has the ability to explore complex topics and conduct intensive research about them, answering very specific questions the user might have. After scouring the internet and analyzing hundreds or even thousands of sources, it’ll present its findings in a comprehensive yet easy-to-read report, the company explains.

The main advantage of Deep Research is that it can do in a matter of minutes what would take a human researcher multiple hours to do, and it can be useful in a range of different business and educational scenarios.

For instance, Google says it’s ideal for industry research, helping businesses to better understand emerging trends they need to be aware of, such as sustainable battery technologies to help inform a company’s clean technology strategies.

Another promising use case is competitive research. An organization could use it to perform a deep analysis of their local and global competitors, or find the ideal location for setting up a new small business, Google said. Alternatively, it can be used for customer research. So a salesperson for example could use it to create a comprehensive report on his or her company’s products and services, tailored for a specific client.

In addition, educators can use Deep Research for tasks such as lesson planning, class project development, presentations and grant writing, creating reports on almost any topic, Google said.

At this time, Deep Research is being made exclusively available to paid Google Workspace users with access to Gemini Advanced. Google Workspace is the company’s productivity suite, containing tools such as Google Docs and Sheets. Gemini Advanced is the premium version of the Gemini AI application, offering support for vastly more tokens than the regular app and additional perks such as more storage space.

To get started with Deep Research in Google Workspace, Gemini Advanced users have to switch to the “1.5 Pro with Deep Research” model in the app, and then enter a simple prompt regarding the topic they wish to research in depth.

Deep Research will then create a multistep research plan that users can revise as needed, and once it’s ready to go, it will immediately start analyzing the most relevant information it finds from its web search. As it works, it will refine its analysis based on what it learns. So, for example, if it’s researching a new trend about clean batteries and discovers a promising new technology in that area, it will automatically dig up as much info as it can about that new tech.

Once the research is done, Deep Research will generate a comprehensive report that summarizes the key findings of what it has learned. According to Google, it’s presented in an easy-to-digest format with relevant links to all of the important, original sources, so users can conduct a deeper dive into different aspects if they need. In addition, users will be able to ask follow-up questions and conduct even deeper research.

Google says Deep Research should be a great time saver, because conducting thorough research is a notoriously time-consuming and monotonous task. It’s also not something that everyone can do, as it requires a certain degree of skill in terms of knowing what to search the web for. According to Google, it can condense hours of research into a matter of minutes on just about any topic.

Deep Research is available now for Google Workspace users on the web, with mobile support coming later. At present, it’s only supported in English, but the company says it will extend it to additional languages soon.

Alongside Deep Research, Google also announced a couple of new experimental models in Gemini Advanced, including 2.0 Flash Thinking Experimental and 2.0 Pro Experimental, which are currently ranked as the world’s second and third most powerful generative AI models respectively on Imarena’s Chatbot Arena LLM Leaderboard.

The 2.0 Flash Thinking Experimental model is Google’s most advanced “reasoning model” thus far. It differs from traditional generative AI models in that it breaks down prompts into a series of steps to improve its reasoning abilities and deliver better responses. Uniquely, it breaks down its entire thought process, so users can trace its line of reasoning and understand the assumptions it made.

Image: News/Dreamina