Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning
By: Yuhao Zhou , Yiheng Wang , Xuming He and more
Potential Business Impact:
Tests AI's ability to do science like a human.
Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.
Similar Papers
Scientific Reasoning: Assessment of Multimodal Generative LLMs
Computation and Language
Helps AI understand science questions better.
Demystifying Scientific Problem-Solving in LLMs by Probing Knowledge and Reasoning
Computation and Language
Helps computers solve science problems better.
SciVideoBench: Benchmarking Scientific Video Reasoning in Large Multimodal Models
CV and Pattern Recognition
Tests AI's ability to understand science videos.