Summary of Connecting the Dots: Evaluating Abstract Reasoning Capabilities Of Llms Using the New York Times Connections Word Game, by Prisha Samadarshi et al.
Connecting the Dots: Evaluating Abstract Reasoning Capabilities of LLMs Using the New York Times Connections Word Game
by Prisha Samadarshi, Mariam Mustafa, Anushka Kulkarni, Raven Rothkopf, Tuhin Chakrabarty, Smaranda Muresan
First submitted to arxiv on: 16 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This research paper evaluates the performance of state-of-the-art large language models (LLMs) against expert and novice human players on the popular word puzzle game, The New York Times Connections. The study uses 438 games to assess the abilities of LLMs, including Claude 3.5 Sonnet, which has shown impressive reasoning skills on various benchmarks. However, even the best-performing LLM can only fully solve 18% of the games, outperformed by both novice and expert human players. The results also highlight the challenges faced by LLMs in categorizing words based on semantic relations, Encyclopedic Knowledge, Multiword Expressions, and combinations of Word Form and Meaning. The study establishes the Connections game as a benchmark for evaluating abstract reasoning capabilities in AI systems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how well big language models do at solving word puzzles. It compares these models to people who are good at puzzles and people who aren’t very good. The models, like Claude 3.5 Sonnet, can only solve some of the puzzles correctly. People who know a lot about words and phrases are actually better at solving the puzzles than the models. This shows that language models still have a long way to go before they’re as good as people at understanding words and meanings. |
Keywords
» Artificial intelligence » Claude
