Google Unveils ERA AI for Scientific Software

Google Research has introduced Empirical Research Assistance (ERA), a new artificial intelligence system designed to write expert-level scientific software ^[1]. The system, which was recently published in the journal Nature, automates the creation of high-performance code for complex research tasks ^[2]. ERA combines large language models with a specialized tree search to achieve expert results ^[1][3].

The technology has already proven successful across diverse fields. Researchers used ERA for tasks in genomics, neuroscience prediction, and forecasting COVID-19 hospitalizations ^[1]. By handling these technical computational requirements, the system significantly speeds up discovery cycles for scientists ^[2][3]. This tool allows researchers to focus more on their primary hypotheses rather than manual coding ^[1].