Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions

arXiv 2025 75.4 method

TLDR

A comprehensive survey of agentic AI systems for scientific discovery, covering progress, challenges, and future directions across multiple scientific fields.

Reasoning

The paper provides a broad, well-structured overview of agentic AI in scientific discovery, with clear categorization and discussion of real-world applications in chemistry, biology, and materials science. However, as a survey, it lacks original experimental results or novel contributions, and the depth of analysis for each sub-area may be limited by the breadth of coverage.

Read-first score

Read-first score 75.4, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 97.

Methodology quality 25%

100

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=analysis,dataset,evaluation,experiment,metric,result

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Topical relevance 42%

80.8

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=dataset

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 22.

Keyword Scores

automated scientific discovery

scientific discovery agent

AI scientist

autonomous research agent

AI for scientific research

research automation

automated research

automated experimentation

literature review agent

experiment design agent

survey generation

paper writing agent

Deep Analysis

Innovations

Comprehensive categorization of Agentic AI systems and tools for scientific discovery across chemistry, biology, and materials science
Discussion of key evaluation metrics, implementation frameworks, and commonly used datasets for Agentic AI
Identification of critical challenges (literature review automation, system reliability, ethical concerns) and future directions emphasizing human-AI collaboration and system calibration

Methodology

The authors conducted a literature survey, categorizing existing Agentic AI systems by scientific domain and capability (literature review, hypothesis generation, experimentation, analysis). They reviewed evaluation metrics, implementation frameworks, and datasets, and synthesized challenges and future research directions.

Key Results

The survey highlights significant progress in Agentic AI applications in chemistry, biology, and materials science, catalogs key evaluation metrics and implementation frameworks, and identifies critical challenges including literature review automation, system reliability, and ethical concerns.

Limitations

Literature review automation remains a significant challenge for current Agentic AI systems
System reliability of Agentic AI in scientific workflows is not yet assured
Ethical concerns surrounding autonomous scientific discovery are unresolved

Resources

Paper