[2603.00883] Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact

Computer Science > Machine Learning arXiv:2603.00883 (cs) [Submitted on 1 Mar 2026] Title:Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact Authors:Michael Hardy, Yunsung Kim View a PDF of the paper titled Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact, by Michael Hardy and Yunsung Kim View PDF HTML (experimental) Abstract:LLMs increasingly excel on AI benchmarks, but doing so does not guarantee validity for downstream tasks. This study evaluates the performance of leading foundation models (FMs, i.e., generative pre-trained base LLMs) with out-of-distribution (OOD) tasks of the teaching and learning of schoolchildren. Across all FMs, inter-model behaviors on disparate tasks correlate higher than they do with expert human behaviors on target tasks. These biases shared across LLMs are poorly aligned with downstream measures of teaching quality and often \textit{negatively aligned with learning outcomes}. Further, we find multi-model ensembles, both unanimous model voting and expert-weighting by benchmark performance, further exacerbate misalignment with learning. We measure that 50\% of the variation in misalignment error is shared across foundation models, suggesting that common pretraining accounts for much of the misalignment in these tasks. We demonstrate methods for robustly measuring alignment of complex tasks and provide unique insights into both educational applications of foundation models and to un...