OpenAI’s new ChatGPT agent is ‘like a superpower,’ says CEO Sam Altman

OpenAI has recently introduced a new AI tool named Deep Research.

This new AI agent is capable of conducting multi-step research on the internet for complex tasks and can accomplish in minutes what would take a human hours.

Simply provide it with a prompt and ChatGPT will search, analyze, and synthesize numerous online sources to generate a comprehensive report comparable to that of a research analyst.

OpenAI showcased this new feature in a video released on Sunday.

Deep Research caters to professionals in various fields like finance, science, policy, and engineering, offering in-depth and reliable insights. It is also useful for consumers looking for personalized recommendations on products that require thorough research, such as cars, appliances, and furniture. The outputs include clear citations and summaries for easy verification. Essentially, this feature streamlines time-consuming research, efficiently delivering niche information with just one query.

In a series of posts shared on X on Sunday, OpenAI’s chief Sam Altman described Deep Research as a “superpower; experts on demand.”

He mentioned that it can conduct complex research and reasoning tasks on the internet and provide a report, handling tasks that would typically take hours or days and cost a significant amount of money.

Although it is computationally intensive and slow, Altman claimed that it is the first AI system capable of performing a wide range of complex and valuable tasks.

Deep Research typically takes between 5 and 30 minutes to complete its work, and users receive a notification once the research is finished. The final output is delivered in the form of a report by ChatGPT. Currently, reports are text-only, but OpenAI plans to add embedded images, data visualizations, and other analytical outputs in the future for better clarity and context.

Deep Research is now available as part of OpenAI’s Pro tier for $200 a month, offering 100 queries per month on the web (with mobile and desktop apps coming by the end of February). It will also be accessible to Plus, Team, and Enterprise customers soon, before eventually being included in OpenAI’s free tier.

Altman encouraged users to test Deep Research on their most challenging tasks that can be solved using the internet to see its capabilities.

However, OpenAI cautioned that the new tool may struggle with distinguishing authoritative information from rumors and currently lacks accuracy in conveying uncertainty. It also mentioned that users may encounter minor formatting errors and delays in tasks, but these issues are expected to improve with more usage and time.

Deep Research follows the recent unveiling of another AI agent by OpenAI called Operator, which acts as a human assistant for performing web-based tasks like making reservations, booking trips, and ordering groceries according to user requests.