Reproduce “Making New Connections: LLMs as Puzzle Generators for The New York Times' Connections Word Game” 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

Tim Merino, Sam Earle, Ryan Sudhakaran, Shyam Sudhakaran, Julian Togelius

The system uses GPT-4 to generate Connections puzzles by building word groups iteratively and introducing intentional overlaps or plausible false groups. Human players evaluated the puzzles against published New York Times puzzles in blind comparisons. LLM-generated false-group puzzles were preferred in 42.86% of responses and tied in 14.29%. The small initial study does not show they are generally better.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.