SciPredict: Can LLMs Predict the Outcomes of Scientific Experiments in Natural Sciences?

TL;DR

SciPredict introduces a benchmark to evaluate LLMs' ability to predict scientific experiment outcomes, revealing current models' limitations and emphasizing the need for reliable prediction awareness.

Contribution

The paper presents SciPredict, a new benchmark with 405 tasks across physics, biology, and chemistry, to assess LLMs' predictive accuracy and reliability in scientific experiments.

Findings

Model accuracies are 14-26%, close to human experts' 20%.

Models fail to reliably distinguish between reliable and unreliable predictions.

Human experts' accuracy improves from 5% to 80% as they judge outcomes as predictable.

Abstract

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