Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Connecting the Dots: Evaluating Abstract Reasoning Capabilities of LLMs Using the New York Times Connections Word Game

Prisha Samdarshi, Mariam Mustafa, Anushka Kulkarni, Raven Rothkopf, Tuhin Chakrabarty, Smaranda Muresan


Abstract
The New York Times Connections game has emerged as a popular and challenging pursuit for word puzzle enthusiasts. We collect438 Connections games to evaluate the performance of state-of-the-art large language models (LLMs) against expert and novice humanplayers. Our results show that even the best-performing LLM, Claude 3.5 Sonnet, which has otherwise shown impressive reasoning abilities on a wide variety of benchmarks, can only fully solve 18% of the games. Novice and expert players perform better than Claude 3.5 Sonnet, with expert human players significantly outperforming it. We create a taxonomy of the knowledge types required to successfully cluster and categorize words in the Connections game. We find that while LLMs are decent at categorizing words based on semantic relations they struggle with other types of knowledge such as Encyclopedic Knowledge, Multiword Expressions or knowledge that combines both Word Form and Meaning. Our results establish the New York Times Connections game as a challenging benchmark for evaluating abstract reasoning capabilities in humans and AI systems.
Anthology ID:
2024.emnlp-main.1182
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
21219–21236
Language:
URL:
https://aclanthology.org/2024.emnlp-main.1182/
DOI:
10.18653/v1/2024.emnlp-main.1182
Bibkey:
Cite (ACL):
Prisha Samdarshi, Mariam Mustafa, Anushka Kulkarni, Raven Rothkopf, Tuhin Chakrabarty, and Smaranda Muresan. 2024. Connecting the Dots: Evaluating Abstract Reasoning Capabilities of LLMs Using the New York Times Connections Word Game. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 21219–21236, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Connecting the Dots: Evaluating Abstract Reasoning Capabilities of LLMs Using the New York Times Connections Word Game (Samdarshi et al., EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.1182.pdf
Data:
 2024.emnlp-main.1182.data.zip