OpenAI has introduced a groundbreaking new tool called Deep Research, which can perform intricate research tasks by gathering and synthesizing information from across the internet. This launch comes just days after the company released another A.I. tool capable of shopping for groceries and booking restaurant reservations. The unveiling of Deep Research signals a major leap in the practical applications of A.I., providing users with the ability to conduct in-depth research that would typically take hours or even days to complete.
During a demonstration on YouTube on Sunday, Kevin Weil, OpenAI’s chief product officer, showcased Deep Research’s capabilities to lawmakers and policymakers in Washington. He highlighted how the tool could complete tasks in just five to 30 minutes that would usually take a human anywhere from 30 minutes to 30 days to accomplish.
Deep Research, which leverages A.I. technology referred to as an “A.I. agent,” goes beyond the capabilities of traditional chatbots by utilizing external software and services. This includes everything from ordering food to synthesizing and analyzing vast amounts of information from websites, PDFs, and images across the internet. During the demonstration, Weil requested the tool to compile a detailed report about the physicist Albert Einstein, which included both factual information about his life and career, as well as potential questions that a senator might ask him during a hypothetical congressional hearing where he’s nominated for U.S. Secretary of Energy.
One of the standout features of Deep Research is its ability to conduct recursive searches. The tool performs a search, which then leads to further inquiries, ultimately synthesizing all gathered data into a cohesive report. The reports generated by Deep Research include citations, showing the sources from which information was gathered, offering transparency in its research process. However, Weil acknowledged that the technology can still make errors, including “hallucinations” — instances where the tool fabricates information or misidentifies sources.
Despite these limitations, Weil emphasized that Deep Research could greatly benefit professionals in fields such as finance, science, and law by enabling them to complete complex tasks with greater speed and accuracy. He also noted that the tool could play a key role in accelerating economic growth in the United States.
The Deep Research tool is based on OpenAI’s latest advancements in reasoning technology, known as OpenAI o3. This technology allows the system to analyze and “reason” through tasks, determining the best course of action based on trial and error. As part of its roll-out, Deep Research is available to ChatGPT Pro users, a $200-per-month service offering access to OpenAI’s most advanced tools. The company plans to make it available through other paid services in the future.
Despite its impressive capabilities, Deep Research’s rollout is not without controversy. OpenAI has recently faced legal challenges, including a lawsuit from The New York Times, which accuses the company and its partner, Microsoft, of copyright infringement regarding the use of news content in A.I. systems. Both OpenAI and Microsoft have denied the claims.
