Show, Don't Tell: Evaluating Large Language Models Beyond Textual Understanding with ChildPlay

TL;DR

This paper introduces ChildPlay, a benchmark suite to evaluate large language models' abilities beyond text, focusing on strategic, spatial, and chemical reasoning through simple ASCII-encoded games, revealing limited generalization.

Contribution

The paper presents a new benchmark set and evaluation methodology for assessing GPT models on non-linguistic tasks like strategy and spatial reasoning, highlighting their limitations.

Findings

GPT models perform poorly on strategic and spatial tasks.

Performance improves with larger models on some tasks.

Models struggle with chemistry-related ASCII graph interpretation.

Abstract

We developed a benchmark set to assess the generalization of state-of-the-art large language models on problems beyond linguistic tasks and evaluate it on a systematic progression of GPT models (GPT-3.5, GPT-4, GPT-4o, GPT-4o-mini). Using simple games like Tic-Tac-Toe, Connect Four, Battleship, and a Shape Recognition Game, all encoded in ASCII, we test strategic capabilities and spatial reasoning, core abilities any artificial intelligence would need to master for solving problems in chemistry. To probe generalization, we introduce two new games for spatial logic: LEGO Connect Language (LCL) and Guess-the-SMILES (GtS), a operationally simple chemistry benchmark. Our results show that GPT models provide meaningful responses for several tasks but, generally, perform poorly. A systematic performance progression with increased model capabilities (GPT-3.5, GPT-4, GPT-4o) is only observed for 4 out of the 7 benchmark tasks. All models consistently struggle with Battleship, LCL, and GtS. This suggests that while GPT models can emulate conversational proficiency and basic rule comprehension, they have limited generalization with respect to strategy and spatial reasoning. Particularly poor performance is observed for interpreting molecular graphs when encoded in ASCII. The results provided by our open-source benchmark suite (\href{https://github.com/BlueVelvetSackOfGoldPotatoes/child-play}{\texttt{ChildPlay} GitHub Repository}) caution against claims of emergent intelligence in GPT models, which appear more specialized than general.