(Reuters) – The prominent artificial intelligence company, OpenAI, introduced a groundbreaking AI tool named “deep research” on Sunday. This tool is designed to conduct intricate research tasks by navigating through multiple stages on the internet.
Deep research utilizes a modified version of the upcoming OpenAI o3 model, specifically crafted for web exploration and data processing. Users are required to provide a prompt, and OpenAI’s chatbot, ChatGPT, will then scour various online sources including text, images, and PDFs. Subsequently, it synthesizes this information to produce a comprehensive report akin to that of a research analyst, as stated by OpenAI.
OpenAI highlighted the efficiency of deep research by stating that it can accomplish tasks in a matter of minutes that would typically take a human several hours. However, the company also acknowledged that deep research is still undergoing development and has some limitations. It may struggle with distinguishing authoritative information from rumors, and there are concerns regarding confidence calibration as it often inaccurately conveys uncertainties.
OpenAI disclosed that deep research is currently accessible on the web version of ChatGPT starting from Sunday and is scheduled to be implemented on mobile and desktop applications by February.
This marks the second AI agent launch by OpenAI this year, following the preview of a tool named Operator in January. Operator offers versatility in tasks such as generating to-do lists or assisting with vacation planning. (Reporting by Rishabh Jaiswal in Bengaluru; Editing by Rashmi Aich)
