OpenAI Unveils Deep Research in ChatGPT, an AI Agent Capable of Conducting Multi-Step Research
OpenAI has introduced a groundbreaking new AI agent within ChatGPT, known as Deep Research, which can autonomously perform multi-step research on complex topics. This new feature, designed for research-intensive tasks, allows ChatGPT Pro users to access an advanced tool capable of conducting in-depth investigations across the internet. Powered by one of OpenAI’s o3 AI models, Deep Research promises to significantly streamline the process of conducting detailed research, which would typically take humans hours, completing it in just a fraction of that time. This addition marks a significant leap forward in AI’s ability to assist with high-level research tasks.
Deep Research is tailored to users engaged in intensive research, especially those in fields like finance, science, policy, and engineering. The AI agent can be used for projects that require comprehensive exploration of various sources, helping users gather the necessary data more efficiently. Additionally, it can be useful for those looking for highly personalized recommendations for large-scale purchases such as cars, furniture, or home appliances, which often involve substantial research. This tool is designed to be an invaluable asset for individuals and professionals who require in-depth, accurate information across a range of industries.
The Deep Research feature is currently available to ChatGPT Pro users, appearing as a button beneath the text field next to the Search icon. To activate it, users simply need to type their query and click on the icon. Once initiated, the AI agent will engage with the user by asking follow-up questions to clarify preferences or gather further details about the research requirements. Users can even upload files to provide additional context, ensuring the research is as relevant and personalized as possible.
Once the agent receives the necessary information, it will begin the research process, and users can track its progress through a thin bar indicating the completion. A sidebar will display a summary of the steps taken and the sources used during the research. Depending on the complexity of the task, the research output can take anywhere from five to 30 minutes to compile, culminating in a comprehensive report that delivers the results in an easy-to-understand format. This new capability positions ChatGPT as a powerful tool for those in need of deep, methodical research assistance.