#How close is AI to automating its own development process?
Understanding how close we are to AI automating its own development process is crucial for researchers, investors, and policymakers alike. Epoch AI, a research organization focused on trends and predictions in artificial intelligence, has developed a framework designed to assess this question with scientific rigor.
The framework identifies over 60 specific tasks integral to AI research and development, and it evaluates these tasks against current automation technologies.
#What does the new taxonomy reveal about AI tasks?
The taxonomy directly leverages data from O*NET, the U.S. Department of Labor's comprehensive catalog of job descriptions across the nation. This database breaks down various occupations into detailed tasks, skills, and work activities. Epoch AI builds upon its earlier findings, which categorized AI research and development into six broad areas: hypothesis creation, experiment design, execution, analysis, communication, and studying prior work. In the new framework, these categories have been dissected into over 60 discrete tasks, reflecting the daily realities faced by professionals in the AI field.
The earlier 2024 study highlighted coding and debugging as areas ripe for automation, signifying that these tasks could be among the first to experience significant technological advancements.
#Why is categorizing tasks in AI important for the future?
This taxonomy is not just a theoretical exercise. Epoch AI is also developing supportive tools to enhance its insights. By early 2026, the organization introduced the GATE macroeconomic model to evaluate how AI automation impacts the broader economy. Previous analyses utilized O*NET data to assess potential automation in remote work and specific research occupations.
One of the leading researchers at Epoch AI, David Owen, has been active in conducting in-depth studies of AI development workflows, employing an O*NET-inspired task labeling system. This method allows for a detailed overview of automation trends, tailored more specifically to the context of AI research and development. Understanding these trends provides valuable insights for individuals and organizations investing in this evolving landscape.