Language Model Goal Selection Differs from Humans' in a Self-Directed Learning Task

TL;DR

This study compares human and large language model goal selection in a cognitive science task, revealing significant differences and highlighting the uniqueness of human goal-setting behavior.

Contribution

It provides empirical evidence that current LLMs diverge from human goal selection patterns in self-directed learning tasks.

Findings

Humans explore diverse goals and learn gradually, while models exploit a single solution.

Models show low variability across instances and limited improvement from reasoning techniques.

Findings are consistent across different experimental settings.

Abstract

Whether in agentic workflows, social studies, or chat settings, large language models (LLMs) are increasingly being asked to replace humans in choosing which goals to pursue, rather than completing predefined tasks. However, the assumption that LLMs accurately reflect human preferences for goal setting remains largely untested. We assess the validity of LLMs as proxies for human goal selection in a controlled, self-directed learning task borrowed from cognitive science. Across five models (GPT-5, Gemini 2.5 Pro, Claude Sonnet 4.5, Qwen3 32B, and Centaur), we find substantial divergence from human behavior. While people gradually explore and learn to achieve goals with diversity across individuals, most models exploit a single identified solution or show surprisingly low performance, with distinct patterns across models and little variability across instances of the same model. Chain-of-thought reasoning and persona steering provide limited improvements, and our conclusions hold across experimental settings. While they await confirmation in applied settings, these findings highlight the uniqueness of human goal selection and caution against its replacement with current models.