OpenAI Unveils AI Research Tool After DeepSeek Disruption: 'Minutes, Not Hours'

OpenAI has unveiled a new AI tool aimed at assisting with research, just as DeepSeek—a fast-growing Chinese AI company known for its cost-effective, high-performance technology—gained attention with its R1 chatbot.

On 2nd February, OpenAI released Deep Research, an AI tool designed to tackle complex research tasks by conducting multi-stage internet searches. It operates on a specialised version of OpenAI's upcoming o3 model, fine-tuned for web browsing and data analysis.

OpenAI's New Research Assistant

Users provide a prompt, and ChatGPT then takes over, locating, examining, and combining information from various online sources, including text, images, and PDFs. According to OpenAI, the result is a thorough report comparable to what a research analyst would produce.

'It accomplishes in tens of minutes what would take a human many hours,' the Sam Altman-led artificial intelligence company said. However, OpenAI noted that 'deep Research is still in development and has some limitations.

'It may struggle with distinguishing authoritative information from rumours and currently shows weakness in confidence calibration, often failing to convey uncertainty accurately,' the company said. Deep Research became available on the web version of ChatGPT on 2nd February, with mobile and desktop app releases planned for later this month.

Deep Research is the second AI agent OpenAI has launched this year. In January, they previewed 'Operator,' a tool designed to handle various tasks, from creating to-do lists to helping with vacation planning.

How Deep Research Works

Deep Research learned its skills through end-to-end reinforcement learning, tackling challenging web browsing and reasoning tasks across diverse fields. This training taught it to plan and conduct multi-step searches for the necessary data, including the ability to backtrack and adapt to real-time information.

The model can also analyse user-uploaded files, create and refine graphs using a Python tool, incorporate generated graphs and images from websites into its responses, and cite specific sentences or passages from its sources. This training has improved performance on several public tests focused on real-world problems.

OpenAI Rethinks Strategy Amidst Rising Competition

The release of Deep Research underscores OpenAI's ongoing innovation. Still, behind the scenes, Altman has expressed concerns about the company's current direction, suggesting that a new open-source approach may be necessary in the wake of DeepSeek's impressive debut.

On 31st January, Altman stated that his prominent AI company is 'on the wrong side of history' regarding transparency about its technology's inner workings. The top executive made these remarks during a Reddit 'Ask Me Anything' session, during which he answered questions, including one about whether OpenAI would consider publishing its Research.

Altman responded that he supported the idea and that it was a subject of ongoing discussion within San Francisco-based OpenAI. 'I personally think we have been on the wrong side of history here and need to figure out a different open source strategy,' he said.

'Not everyone at OpenAI shares this view, and it's also not our current highest priority,' Altman added. The Chinese AI startup DeepSeek has garnered attention not only for the reportedly low cost and high performance of its R1 chatbot but also for its claim to be a publicly beneficial "open-source" project, contrasting itself with the closed models offered by OpenAI and Google.

Altman Signals Potential Shift For OpenAI

Open source alludes to programmers making the source code of their software available rather than just the compiled, ready-to-run program. According to a report by Fortune, this approach has often conflicted with private companies seeking revenue and intellectual property protection.

Meta, DeepSeek, and the French AI developer Mistral differentiate themselves by providing developers with free access to their tools' underlying code. A Reddit user asked Altman if DeepSeek had influenced his plans for future OpenAI models.

'It's a very good model,' Altman said of DeepSeek. 'We will produce better models, but we will maintain less of a lead than we did in previous years.'