Recently, OpenAI introduced a new artificial intelligence tool called "Deep Research." It aims to assist users in efficiently completing various online research tasks, covering multiple areas from complex scientific issues to car recommendations. The tool thoroughly explores online resources and provides users with detailed research reports, greatly enhancing research efficiency.
Compared to traditional AI assistants, Deep Research can complete tasks in as little as 5 to 30 minutes. Users can assign tasks to the AI while they attend to other matters. The final results are presented in a report format, and soon it will offer embedded images, data visualization, and other analytical outputs to improve clarity and readability.
In OpenAI's "The Final Exam for Humanity" challenge, Deep Research performed excellently, achieving an impressive 26.6% accuracy rate, far surpassing its competitor DeepSeek R1's 10% accuracy. This result highlights the iterative optimization and powerful comprehensive capabilities of Deep Research in the rapidly evolving field of artificial intelligence, demonstrating that while some tasks require time, AI shows great potential in solving them.
OpenAI announced in its official blog that the Deep Research feature will be available to specific paying customers through ChatGPT. Users can request AI to conduct online searches of text, images, PDF files, and uploaded documents to generate detailed reports. OpenAI compares its functionality to that of a professional research analyst, emphasizing that this tool can complete tasks in "tens of minutes" that would typically take humans several hours.
However, OpenAI also reminds users that Deep Research is still in its early stages and may face challenges in identifying misinformation and accurately distinguishing rumors. Furthermore, the tool consumes a significant amount of computing resources, so initially, users can submit a maximum of 100 queries per month.
