A week ago, OpenAI released a tool that can go online to shop for groceries or book a restaurant reservation. Now it is offering A.I. technology that can gather information from across the internet and synthesize it in concise reports.
OpenAI unveiled the new tool, called Deep Research, with a demonstration on YouTube on Sunday, days after showing the technology to lawmakers, policymakers and other officials in Washington.
“It can do complex research tasks that might take a person anywhere from 30 minutes to 30 days,” Kevin Weil, OpenAI’s chief product officer, said at the event in Washington. By contrast, Deep Research can accomplish such tasks in five to 30 minutes, depending on the complexity.
Artificial intelligence researchers call this kind of technology an A.I. agent. While chatbots can answer questions, write poems and generate images, agents can use other software and services on the internet. This might involve anything from ordering dinner via DoorDash to synthesizing information from across the internet.
During the briefing on Capitol Hill, Mr. Weil showed the technology gathering information about Albert Einstein. He asked the tool to put together a detailed report about the physicist for a hypothetical Senate staff member preparing for a congressional hearing where Einstein is a nominee for U.S. secretary of energy.
In addition to providing information about Einstein’s background and personality, it generated five questions that a senator could ask the physicist to determine whether he was the right person for the job.
“It can surf the web and understand text and images and PDFs,” Mr. Weil said. “And it can do this recursively. It can do one search, and that leads to other searches, and then it can synthesize all the information it has learned.”
Mr. Weil said the reports generated by the tool included citations showing where the information was found. But A.I. technologies like this can still get things wrong or even make up information — a phenomenon that A.I. researchers call “hallucination.” This may mean that it provides incorrect citations.
OpenAI said that the tool might struggle to distinguish authoritative information from rumors and that it often failed to accurately convey when it was uncertain about the information it was delivering.
Still, Mr. Weil argued that the tool could help the United States accelerate economic growth. He added that the tool would be particularly useful for people in fields like finance, science and law.
(The New York Times has sued OpenAI and its partner, Microsoft, accusing them of copyright infringement of news content related to A.I. systems. OpenAI and Microsoft have denied those claims.)
OpenAI said that, beginning on Sunday, Deep Research would be available to anyone who was subscribed to ChatGPT Pro, a $200-a-month service that provides access to all of the company’s latest tools. It plans to also offer the tool via its other paid services.
The tool is based on the same technology that drives ChatGPT. This technology is what A.I. researchers call a neural network — a mathematical system that can learn skills by analyzing data.
In recent months, OpenAI has developed versions of the technology that can “reason” through tasks, determining through trial and error what actions to take. Deep Research is based on the company’s newest reasoning technology, OpenAI o3.
