OpenAI has laid out a concrete timeline for when it believes artificial intelligence will be capable of doing real scientific research, not just summarizing papers or answering questions, but actually designing experiments, running analyses, and proposing next steps. The company expects to have an intern-level AI research assistant operational by September 2026, with a fully autonomous AI researcher arriving by March 2028.
What “intern-level” actually means
The intern-level AI isn’t just expected to answer prompts or generate summaries. OpenAI envisions a system capable of performing end-to-end research, meaning it could take a research question, design an appropriate methodology, execute the analysis, interpret results, and then propose follow-up experiments.
Jakub Pachocki, OpenAI’s Chief Scientist, has indicated that deep learning systems may reach superintelligence within a decade. These goals were publicly announced during a livestream featuring CEO Sam Altman and Pachocki on October 28.
The $1.4 trillion question
