Reliable Evaluation of Large Language Models
Develop reliable evaluation methodologies for large language models that effectively assess model performance, including helpfulness and harmlessness, addressing the acknowledged unresolved problem of evaluating such models.
References
However, evaluating LLMs has consistently been a challenging and unresolved problem.
Second, declared evidence markers give verifiable ground truth for an engagement whose author can state them, and open-ended harm still depends on a model's judgement.
Some benchmarks include human studies, but these remain contentious, as designing sound human-model evaluations is an open question and a direction for future work.
The implementation running on a device, with its leakage, faults, and countermeasures, remains open.
The cryptographic module inside the device remains open as an object of evaluation.
However, open questions remain regarding stochastic variability, error behavior, and cross-model performance.
As a result, despite growing interest in deployment, robust evaluation of clinical information retrieval remains an open challenge [16, 17].
Its reliability for assessing LLM outputs remains to be established.
The open frontier is outputs that execute cleanly yet are semantically wrong.
Evaluation on general-purpose frontier LLMs as judges is left to future work.
Our LLM-as-judge achieves 100\% agreement with human evaluation on 200 sampled cases, but some errors may still exist in the remaining samples. We will further explore more robust and programmatic evaluation procedures to improve evaluation accuracy and reliability.
Inter-LLM disagreement may therefore be a useful label-free signal for which phenotypes most need prompt refinement or human review, although the present spot-check is too small to validate it.
As agents move toward real-world deployment, how to evaluate them and when to trust them remain open challenges~\citep{kapoor2025agents, kirchhof2025position}.