Home Cloud Computing OpenAI Launches “Deep Research” in ChatGPT: A New AI Agent for Complex...

OpenAI Launches “Deep Research” in ChatGPT: A New AI Agent for Complex Web Research

85
0

OpenAI has unveiled a groundbreaking new feature in ChatGPT called “deep research,” an AI agent designed to conduct multi-step research on the internet for complex tasks.

The company claims this new capability can accomplish in tens of minutes what would typically take a human many hours.

Deep research is powered by a version of OpenAI’s upcoming o3 model, optimized for web browsing and data analysis.

It leverages reasoning skills to search, interpret, and analyze vast amounts of text, images, and PDFs online, adapting its search strategy as it encounters new information.

“The ability to synthesize knowledge is a prerequisite for creating new knowledge,” OpenAI stated, highlighting the significance of this development toward their broader goal of achieving Artificial General Intelligence (AGI).

This new tool is aimed at professionals engaged in intensive knowledge work across fields like finance, science, policy, and engineering, requiring thorough, precise, and reliable research.

OpenAI also suggests it could be valuable for consumers making complex purchasing decisions, such as buying cars, appliances, or furniture.

A key feature of deep research is its comprehensive documentation, providing clear citations and a summary of its reasoning process, enabling users to easily reference and verify the information.

It excels at uncovering niche, non-intuitive information often hidden across numerous websites.

“Deep research was trained using real-world tasks that required browser and Python tool use, employing the same reinforcement learning methods used for OpenAI o1, their first reasoning model. While o1 demonstrated proficiency in coding, math, and other technical areas, deep research addresses the need for extensive context and information gathering from diverse online sources, bridging the gap for real-world challenges.”

Users can access deep research by selecting the option in the ChatGPT message composer and entering their query.

They can specify their needs, whether it’s a competitive analysis or personalized recommendations, and even attach files or spreadsheets for added context.

A sidebar provides a summary of the steps taken and sources used during the research process.

The research duration can range from 5 to 30 minutes, allowing the AI to thoroughly explore the web.

Users receive a notification upon completion, and the final output is presented as a report within the chat.

OpenAI noted it plans to add embedded images, data visualizations, and other analytical outputs to these reports in the coming weeks.

OpenAI differentiates deep research from GPT-4o, which is better suited for real-time, multimodal conversations.

Deep research, with its capacity for extensive exploration and citation of each claim, is designed for multi-faceted, domain-specific inquiries where depth and detail are paramount.

It provides a well-documented, verified answer suitable for use as a work product, going beyond a quick summary.

LEAVE A REPLY

Please enter your comment!
Please enter your name here