SciPredict: Can LLMs Predict the Outcomes of Scientific Experiments in Natural Sciences?
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913025165885440 |
|---|---|
| author | Sehwag, Udari Madhushani Lau, Elaine Oskouie, Haniyeh Ehsani Shabihi, Shayan Liang, Erich Toledo, Andrea Mangialardi, Guillermo Fonrouge, Sergio Cardona, Ed-Yeremai Hernandez Vergara, Paula Tyagi, Utkarsh Zhang, Chen Bo Calvin Bhatter, Pavi Johnson, Nicholas Huang, Furong Montoya, Ernesto Gabriel Hernandez Liu, Bing |
| author_facet | Sehwag, Udari Madhushani Lau, Elaine Oskouie, Haniyeh Ehsani Shabihi, Shayan Liang, Erich Toledo, Andrea Mangialardi, Guillermo Fonrouge, Sergio Cardona, Ed-Yeremai Hernandez Vergara, Paula Tyagi, Utkarsh Zhang, Chen Bo Calvin Bhatter, Pavi Johnson, Nicholas Huang, Furong Montoya, Ernesto Gabriel Hernandez Liu, Bing |
| contents | Accelerating scientific discovery requires the identification of which experiments would yield the best outcomes before committing resources to costly physical validation. While existing benchmarks evaluate LLMs on scientific knowledge and reasoning, their ability to predict experimental outcomes - a task where AI could significantly exceed human capabilities - remains largely underexplored. We introduce SciPredict, a benchmark comprising 405 tasks derived from recent empirical studies in 33 specialized sub-fields of physics, biology, and chemistry. SciPredict addresses two critical questions: (a) can LLMs predict the outcome of scientific experiments with sufficient accuracy? and (b) can such predictions be reliably used in the scientific research process? Evaluations reveal fundamental limitations on both fronts. Model accuracies are 14-26% and human expert performance is $\approx$20%. Although some frontier models exceed human performance model accuracy is still far below what would enable reliable experimental guidance. Even within the limited performance, models fail to distinguish reliable predictions from unreliable ones, achieving only $\approx$20% accuracy regardless of their confidence or whether they judge outcomes as predictable without physical experimentation. Human experts, in contrast, demonstrate strong calibration: their accuracy increases from $\approx$5% to $\approx$80% as they deem outcomes more predictable without conducting the experiment. SciPredict establishes a rigorous framework demonstrating that superhuman performance in experimental science requires not just better predictions, but better awareness of prediction reliability. For reproducibility all our data and code are provided at https://github.com/scaleapi/scipredict |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_10718 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SciPredict: Can LLMs Predict the Outcomes of Scientific Experiments in Natural Sciences? Sehwag, Udari Madhushani Lau, Elaine Oskouie, Haniyeh Ehsani Shabihi, Shayan Liang, Erich Toledo, Andrea Mangialardi, Guillermo Fonrouge, Sergio Cardona, Ed-Yeremai Hernandez Vergara, Paula Tyagi, Utkarsh Zhang, Chen Bo Calvin Bhatter, Pavi Johnson, Nicholas Huang, Furong Montoya, Ernesto Gabriel Hernandez Liu, Bing Artificial Intelligence Accelerating scientific discovery requires the identification of which experiments would yield the best outcomes before committing resources to costly physical validation. While existing benchmarks evaluate LLMs on scientific knowledge and reasoning, their ability to predict experimental outcomes - a task where AI could significantly exceed human capabilities - remains largely underexplored. We introduce SciPredict, a benchmark comprising 405 tasks derived from recent empirical studies in 33 specialized sub-fields of physics, biology, and chemistry. SciPredict addresses two critical questions: (a) can LLMs predict the outcome of scientific experiments with sufficient accuracy? and (b) can such predictions be reliably used in the scientific research process? Evaluations reveal fundamental limitations on both fronts. Model accuracies are 14-26% and human expert performance is $\approx$20%. Although some frontier models exceed human performance model accuracy is still far below what would enable reliable experimental guidance. Even within the limited performance, models fail to distinguish reliable predictions from unreliable ones, achieving only $\approx$20% accuracy regardless of their confidence or whether they judge outcomes as predictable without physical experimentation. Human experts, in contrast, demonstrate strong calibration: their accuracy increases from $\approx$5% to $\approx$80% as they deem outcomes more predictable without conducting the experiment. SciPredict establishes a rigorous framework demonstrating that superhuman performance in experimental science requires not just better predictions, but better awareness of prediction reliability. For reproducibility all our data and code are provided at https://github.com/scaleapi/scipredict |
| title | SciPredict: Can LLMs Predict the Outcomes of Scientific Experiments in Natural Sciences? |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.10718 |