ERA AI Generates 40 Single‑Cell Methods, Beats CDC
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- ERA combines a large language model with a tree‑search algorithm to iteratively improve a quality metric and explore the space of possible scientific software solutions.
- ERA discovered 40 novel single‑cell analysis methods that outperformed the top human‑crafted methods on a public leaderboard.
- ERA generated 14 COVID‑19 hospitalization‑forecasting models that outperformed the CDC ensemble and every individual competitor model.
- ERA also produced expert‑level code for geospatial analysis, zebrafish neural‑activity prediction, numerical integral solving, and a new rule‑based time‑series forecasting approach.
- Google DeepMind led the development of ERA with contributions from Google Research, MIT, Harvard and other institutions.
Why it matters: Scientists now have an AI that produced 40 novel single‑cell analysis methods and 14 COVID‑19 forecasting models, cutting the time needed to write expert‑level code and speeding up research breakthroughs across biology and epidemiology. The system’s tree‑search approach demonstrates a scalable way to explore complex scientific problems, a capability applicable across computational science.
Ask SkimNews
