This morning, OpenAI suddenly held a "surprise" live broadcast. During the live broadcast, OpenAI announced, “We are launching a feature called Deep Research, which is a multi-step research model that can discover content, integrate content, and reason about this content. It will even pose “clarifying” questions regarding your prompts to ensure that its multi-step research stays on track. ”
According to reports, this in-depth research feature will be launched in the professional version of ChatGPT later today and will gradually be applied to other OpenAI products. Moreover, OpenAI CEO Sam Altman revealed that free users can also get a very small amount of usage credits.
Launched in a Hurry, three times better than DeepSeek?
Deep Research is a new intelligent agent capability that can conduct multi-step research on the Internet for complex tasks, and can complete in dozens of minutes what would take humans hours to complete. Just give a prompt and ChatGPT will find, analyze, and synthesize hundreds of online information sources to produce a comprehensive report that's as good as a research analyst's.
OpenAI emphasizes that the deep research function is built for those who are engaged in high-intensity knowledge work in fields such as finance, science, policy, engineering, and require in-depth and reliable research. This feature is also useful for savvy consumers looking for highly personalized recommendations when purchasing items that require careful research.
In the coming weeks, OpenAI will also add embedded images, data visualizations, and other analysis output to analysis reports generated by the Deep Research feature to further improve clarity and provide more context. OpenAI sees the deep research capabilities as a major step toward its ambitious goal of developing artificial general intelligence (AGI).
Capability refreshes the highest score in the world and is well received by professional users
In actual tests, many professional users expressed their affirmation of this in-depth research function of OpenAI. Ethan Mollick, a professor at Wharton School who studies artificial intelligence, compared it with Google Gemini and said that Google's similar function only aggregates many information sources, while OpenAI's function is more like letting a person with unique insights (often almost at the Ph.D. level) researchers to follow the clues.
Kevin A. Bryan, associate professor of strategic management at the Rotman School of the University of Toronto and co-founder of AllDayTA, wrote the first draft of an economics paper using the method. His evaluation of the generated content was: “How far can it achieve with just one instruction? To be honest, it’s not bad. To be honest, some of the papers I’ve reviewed are not as good as this. The path from this step to significantly accelerating research progress is quite clear.”
Will top AI manufacturers join the open source competition?
Regarding the release of OpenAI, some netizens commented, "Imagine an open source product with a valuation of billions of dollars, but it meets 95% of customers' needs for free. At this time, everyone must be mobilized to prevent investors from completely panic is out of control. This situation is like rowing, it’s not easy to deal with.”
Some netizens thought, "This is a bit like the competition between Linux and Windows in the 1990s. I guess it won't be long before OpenAI, Gemini, Meta and Grok will launch lightweight open source engines for enthusiasts to play with. In fact, Grok's engine has been open sourced.”
