Reproduce “Application of AI to formal methods - an analysis of current trends” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sebastian Stock, Jannik Dunkelau, Atif Mashkoor
This systematic mapping study quantitatively analyzes 189 primary research papers from 2019-2023, providing a comprehensive overview of AI applications in Formal Methods (FM). The study reveals Neural Networks are predominantly used as an assistance tool in Theorem Proving and SAT solving, while highlighting a significant scarcity of unified datasets and benchmarks.
Key insights
- The field of AI for Formal Methods is nascent, characterized by a high proportion of conference papers and methodologies, with a notable scarcity of standardized benchmarks and case studies.
- AI predominantly serves as an 'assistance tool' to enhance existing FM processes (e.g., finding proofs or clauses) rather than replacing FM tools with AI-generated, potentially non-deterministic, guarantees.
- A critical lack of publicly available, high-quality, and unified datasets and benchmarks significantly hinders reproducibility and comparability of research findings in this domain.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.