OpenAI revealed plans to automate AI research, aiming to deploy an AI research intern by 2025 and fully autonomous AI researchers by 2028, which could dramatically accelerate scientific discovery through recursive self-improvement. This roadmap reflects a broader industry trend toward automating AI development, potentially leading to superintelligent systems within the next decade and transforming the pace of technological advancement.
OpenAI recently held a live stream where they discussed their internal roadmap and goals for AI development, particularly focusing on the automation of AI research. Sam Altman and Yakob Pachokei shared insights into their vision for the near future, revealing timelines that had not been publicly disclosed before. They anticipate that by September 2025, an automated AI research intern will be operational, capable of assisting human researchers. By March 2028, they expect to have fully autonomous AI systems conducting AI research independently. This marks a significant milestone in AI development, as automating AI research itself could dramatically accelerate progress in the field.
The importance of automating AI research lies in its potential to exponentially increase the pace of scientific discovery and technological advancement. Currently, AI research is predominantly conducted by human experts, and while progress has been rapid—especially since 2019 when large language models began to emerge—there are inherent limits to human capacity. Automating research means AI systems could recursively improve themselves, leading to faster breakthroughs and potentially triggering an intelligence explosion. This concept, often referred to as recursive self-improvement, suggests that once AI reaches a certain level of capability, it could rapidly surpass human intelligence.
OpenAI’s roadmap aligns with developments seen at other leading AI organizations like Google DeepMind. DeepMind has already demonstrated early forms of recursive self-improvement with projects such as AlphaFold and other AI systems that optimize hardware and training processes. These advancements have improved efficiency and performance in their data centers and AI models. Although the exact details of DeepMind’s internal plans are not public, it is likely they have similar timelines and goals for automating AI research, indicating a broader industry trend toward this transformative technology.
Yakob Pachokei emphasized that OpenAI’s research program is centered on understanding deep learning and its scaling effects, with a particular focus on artificial general intelligence (AGI). They believe that superintelligent systems—AI smarter than humans across many critical domains—could emerge within the next decade. A key impact of such technology would be the acceleration of scientific discovery, fundamentally changing how quickly new technologies are developed. OpenAI measures progress by how much time AI models save compared to humans in solving complex problems, and they see significant room for improvement as models continue to scale and innovate.
In summary, OpenAI’s disclosure of their internal roadmap highlights a pivotal moment in AI development. The transition from human-led AI research to automated AI research systems could lead to unprecedented advancements and an intelligence explosion. This shift is expected to occur within the next few years, with an AI research intern by 2025 and fully autonomous AI researchers by 2028. The implications are profound, as this could accelerate scientific and technological progress at a scale never seen before, reshaping the future of AI and its role in society.