An AI system to help scientists write expert-level empirical software
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914583839506432 |
|---|---|
| author | Aygün, Eser Belyaeva, Anastasiya Comanici, Gheorghe Coram, Marc Cui, Hao Garrison, Jake Kast, Renee Johnston Anton McLean, Cory Y. Norgaard, Peter Shamsi, Zahra Smalling, David Thompson, James Venugopalan, Subhashini Williams, Brian P. He, Chujun Martinson, Sarah Plomecka, Martyna Wei, Lai Zhou, Yuchen Zhu, Qian-Ze Abraham, Matthew Brand, Erica Bulanova, Anna Cardille, Jeffrey A. Co, Chris Ellsworth, Scott Joseph, Grace Kane, Malcolm Krueger, Ryan Kartiwa, Johan Liebling, Dan Lueckmann, Jan-Matthis Raccuglia, Paul Xuefei Wang Chou, Katherine Manyika, James Matias, Yossi Platt, John C. Dorfman, Lizzie Mourad, Shibl Brenner, Michael P. |
| author_facet | Aygün, Eser Belyaeva, Anastasiya Comanici, Gheorghe Coram, Marc Cui, Hao Garrison, Jake Kast, Renee Johnston Anton McLean, Cory Y. Norgaard, Peter Shamsi, Zahra Smalling, David Thompson, James Venugopalan, Subhashini Williams, Brian P. He, Chujun Martinson, Sarah Plomecka, Martyna Wei, Lai Zhou, Yuchen Zhu, Qian-Ze Abraham, Matthew Brand, Erica Bulanova, Anna Cardille, Jeffrey A. Co, Chris Ellsworth, Scott Joseph, Grace Kane, Malcolm Krueger, Ryan Kartiwa, Johan Liebling, Dan Lueckmann, Jan-Matthis Raccuglia, Paul Xuefei Wang Chou, Katherine Manyika, James Matias, Yossi Platt, John C. Dorfman, Lizzie Mourad, Shibl Brenner, Michael P. |
| contents | The cycle of scientific discovery is frequently bottlenecked by the slow, manual creation of software to support computational experiments\cite{hannay2009how}. To address this, we present Empirical Research Assistance (ERA), an AI system that creates expert-level scientific software whose goal is to maximize a quality metric. The system uses a Large Language Model (LLM) and Tree Search (TS)\cite{silver2016mastering} to systematically improve the quality metric and intelligently navigate the large space of possible solutions. ERA achieves expert-level results when it explores and integrates complex research ideas from external sources. The effectiveness of tree search is demonstrated across a diverse range of tasks. In bioinformatics, ERA discovered 40 novel methods for single-cell data analysis that outperformed the top human-developed methods on a public leaderboard. In epidemiology, ERA generated 14 models that outperformed the CDC ensemble and all other individual models for forecasting COVID-19 hospitalizations. ERA also produced expert-level software for geospatial analysis, neural activity prediction in zebrafish, and numerical solution of integrals, and a novel rule-based construction for time series forecasting. By devising and implementing novel solutions to diverse tasks, ERA represents a significant step towards accelerating scientific progress. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_06503 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | An AI system to help scientists write expert-level empirical software Aygün, Eser Belyaeva, Anastasiya Comanici, Gheorghe Coram, Marc Cui, Hao Garrison, Jake Kast, Renee Johnston Anton McLean, Cory Y. Norgaard, Peter Shamsi, Zahra Smalling, David Thompson, James Venugopalan, Subhashini Williams, Brian P. He, Chujun Martinson, Sarah Plomecka, Martyna Wei, Lai Zhou, Yuchen Zhu, Qian-Ze Abraham, Matthew Brand, Erica Bulanova, Anna Cardille, Jeffrey A. Co, Chris Ellsworth, Scott Joseph, Grace Kane, Malcolm Krueger, Ryan Kartiwa, Johan Liebling, Dan Lueckmann, Jan-Matthis Raccuglia, Paul Xuefei Wang Chou, Katherine Manyika, James Matias, Yossi Platt, John C. Dorfman, Lizzie Mourad, Shibl Brenner, Michael P. Artificial Intelligence Quantitative Methods The cycle of scientific discovery is frequently bottlenecked by the slow, manual creation of software to support computational experiments\cite{hannay2009how}. To address this, we present Empirical Research Assistance (ERA), an AI system that creates expert-level scientific software whose goal is to maximize a quality metric. The system uses a Large Language Model (LLM) and Tree Search (TS)\cite{silver2016mastering} to systematically improve the quality metric and intelligently navigate the large space of possible solutions. ERA achieves expert-level results when it explores and integrates complex research ideas from external sources. The effectiveness of tree search is demonstrated across a diverse range of tasks. In bioinformatics, ERA discovered 40 novel methods for single-cell data analysis that outperformed the top human-developed methods on a public leaderboard. In epidemiology, ERA generated 14 models that outperformed the CDC ensemble and all other individual models for forecasting COVID-19 hospitalizations. ERA also produced expert-level software for geospatial analysis, neural activity prediction in zebrafish, and numerical solution of integrals, and a novel rule-based construction for time series forecasting. By devising and implementing novel solutions to diverse tasks, ERA represents a significant step towards accelerating scientific progress. |
| title | An AI system to help scientists write expert-level empirical software |
| topic | Artificial Intelligence Quantitative Methods |
| url | https://arxiv.org/abs/2509.06503 |