On the Limits of LLM Adaptability: Impact of Model-Internalized Priors on Annotation Task Performance

Abstract:Large Language Models (LLMs) are increasingly used for zero-shot annotation and LLM-as-a-judge tasks, yet their reliability hinges on how model-internalized priors interact with user-provided instructions. We investigate three dimensions of this interaction: (1) how an LLM's familiarity with data and task definitions relates to performance, (2) whether additional information in prompts can correct zero-shot errors ("decision stickiness"), and (3) model susceptibility to misaligned task definitions. We introduce Definition-Specific Familiarity (DSF), which measures alignment between a model's elicited concept and the target definition. Across nine LLMs and six toxicity datasets (five primary datasets plus an additional robustness dataset), DSF predicts annotation performance after controlling for dataset identity (partial $r=+0.41$). This association remains positive across all prompting conditions tested. In contrast, three common text-memorization metrics show no positive association. We show that prompting has limited corrective power: only 34.8% of zero-shot errors are corrected by additional instructions or examples, with high-confidence errors especially persistent. Misaligned definitions systematically shift predictions without reducing reported confidence, making confidence unreliable for detecting definition-policy mismatch. Together, these findings establish definition alignment as a practical model-selection criterion and show that better prompting alone cannot substitute for validating model-policy fit.

Updated based on camera-ready from ICML 2026 (Oral & Spotlight); PMLR vol. 306. 9 pages, 5 figures

Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)

I.2.7; I.2.6; K.4.0

arXiv:2606.00467 [cs.CL]

(or arXiv:2606.00467v2 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2606.00467 Focus to learn more arXiv-issued DOI via DataCite

Proceedings of the 43 rd International Conference on Machine Learning, Seoul, South Korea. PMLR 306, 2026

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.