OpenAI has sent shockwaves through the AI industry with the release of its new Deep Research capability, which allows AI models to conduct extensive internet research and synthesize information over extended periods.
Announced less than 24 hours ago, Deep Research is already being hailed as a potential inflection point by AI leaders.
Emad Mostaque, founder of Stability AI, declared on social media that “We are quite clearly in an intelligence takeoff scenario.” He emphasized that machines will soon be able to perform most digital knowledge tasks faster, better, and cheaper than humans, with the ability to coordinate, scale, and learn from mistakes.
The rapid progress is evident in benchmark results. Kevin Roose, a New York Times tech columnist, noted that just 10 days ago, the leading AI model achieved 8.3% accuracy on Humanity’s Last Exam. Now, Deep Research has set a new high score of 26.6% accuracy. This dramatic improvement in such a short time frame has left many in the field stunned.
Experts attribute this accelerated development in part to the “DeepSeek effect.” While OpenAI pioneered reasoning models with its O1 family, DeepSeek’s open-source release and detailed documentation of their techniques have allowed the entire industry to iterate and improve rapidly.
Deep Research represents a convergence of two major trends in AI development: autonomous agents and powerful reasoning models. By combining these capabilities, OpenAI has created a system that can conduct research with the depth and nuance of human experts, but at machine speed.
Early reports from users are overwhelmingly positive. Dario Unutmaz, a professor and biomedical scientist, described Deep Research as an “absolute game-changer” for scientific research, publishing, legal documents, medicine, and education.
He reported that the system produced “impeccable” reports on cancer cases, comparable to what only specialist MDs could write.
The potential impact on various industries is significant. Sam Altman, CEO of OpenAI, estimates that Deep Research can already accomplish a “single-digit percentage” of all economically valuable work globally – a figure that translates to trillions of dollars.
However, not everyone is celebrating. Some Google employees have pointed out similarities between Deep Research and their own products, including the name itself.
This has led to some public commentary on social media about the origins of the ideas behind Deep Research.
As impressive as Deep Research appears to be, Altman hinted that there are even more significant developments on the horizon. He stated that Deep Research is not even the “one more thing” OpenAI has planned, teasing further announcements in the coming days.
The launch of Deep Research marks a significant milestone in AI development, potentially bringing us closer to artificial general intelligence (AGI). As the technology continues to evolve at a breakneck pace, its implications for research, industry, and society at large are only beginning to be understood.
