Ai2 Paper Finder imitates how researchers actually search: breaking a query down, following citations, evaluating relevance and iterating. Natural-language queries run through LLMs, citation tracking and semantic analysis to find papers — including obscure ones keyword search misses.
Benefits & outcomes
Finds niche or less-cited papers standard engines overlook, with summaries explaining each paper's relevance.
Handles complex multi-criteria queries as written.
Fast mode and exhaustive mode.
On academic benchmarks: perfectly relevant papers for 89% of queries, highly relevant for 98%.
Notes & quotes
"We believe that AI-based literature search should follow the research and thought process that a human researcher would use when looking for relevant papers in their field."
Still a work in progress, by Ai2's own description.
Publisher / company: Allen Institute for AI (Ai2)
Date of mention: 26/03/2025
Source: https://allenai.org/blog/paper-finder