Frontier LLMs Still Struggle with Simple Reasoning Tasks

TL;DR

This paper reveals that state-of-the-art large language models still struggle with simple reasoning tasks, showing systematic failures and poor out-of-distribution generalization even on trivialized problems.

Contribution

The study introduces a suite of procedurally generated simple reasoning tasks and the unpuzzles dataset to analyze model failures and out-of-distribution generalization issues.

Findings

Models fail on simple reasoning tasks due to statistical shortcuts and errors.

Modern LLMs perform poorly on trivialized puzzles, indicating memorization issues.

Out-of-distribution generalization remains a challenge for frontier LLMs.

Abstract

While state-of-the-art large language models (LLMs) demonstrate advanced reasoning capabilities-achieving remarkable performance on challenging competitive math and coding benchmarks-they also frequently fail on tasks that are easy for humans. This work studies the performance of frontier LLMs on a broad set of such "easy" reasoning problems. By extending previous work in the literature, we create a suite of procedurally generated simple reasoning tasks, including counting, first-order logic, proof trees, and travel planning, with changeable parameters (such as document length. or the number of variables in a math problem) that can arbitrarily increase the amount of computation required to produce the answer while preserving the fundamental difficulty. While previous work showed that traditional, non-thinking models can be made to fail on such problems, we demonstrate that even state-of-the-art thinking models consistently fail on such problems and for similar reasons (e.g. statistical shortcuts, errors in intermediate steps, and difficulties in processing long contexts). To further understand the behavior of the models, we introduce the unpuzzles dataset, a different "easy" benchmark consisting of trivialized versions of well-known math and logic puzzles. Interestingly, while modern LLMs excel at solving the original puzzles, they tend to fail on the trivialized versions, exhibiting several systematic failure patterns related to memorizing the originals. We show that this happens even if the models are otherwise able to solve problems with different descriptions but requiring the same logic. Our results highlight that out-of-distribution generalization is still problematic for frontier language models and the new generation of thinking models, even for simple reasoning tasks, and making tasks easier does not necessarily imply improved performance.