Revisiting Epistemic Markers in Confidence Estimation: Can Markers Accurately Reflect Large Language Models' Uncertainty?

Jiayu Liu, Qing Zong, Weiqi Wang, Yangqiu Song

TL;DR

While Large Language Models (LLMs) can use natural language phrases (like "fairly certain" or "probably") to express confidence reliably within a specific domain, this capability breaks down when they face new topics. Out-of-distribution shifts cause the internal meaning and relative ranking of these verbal confidence markers to fluctuate wildly, making them unreliable for general uncertainty estimation.

What Problem Does This Paper Address?

When humans communicate uncertainty, we rarely say, "I am 74% confident in this statement." Instead, we use epistemic markers—phrases like "probably," "I think so," or "undoubtedly." For LLMs to be safely deployed in high-stakes fields like medicine or law, they need to communicate their confidence in these same natural, human-readable ways.

While previous research has focused on whether LLM expressions match human expectations, this paper tackles a more fundamental question: Do LLMs even maintain a stable internal mapping for the words they choose? If a model says "probably" on a math test versus a legal exam, does it actually correspond to the same level of underlying accuracy?

Methods

The authors define marker confidence empirically as the actual observed accuracy of a model's answers when it chooses to use a specific epistemic marker. For example, if a model includes the phrase "fairly certain" in 100 answers, and 65 of those answers are correct, the marker confidence for "fairly certain" is 65%.

Figure 1: An example of our framework calculating the marker confidence of “fairly certain” for GPT-4o on StrategyQA. We calculate the confidence for all the markers across seven models and seven datasets.

To evaluate how stably models use these markers, the authors prompt LLMs to answer questions while explicitly incorporating a single verbal marker. They then extract these markers and evaluate them across seven newly proposed metrics spanning three categories:

  • Calibration Error (ECE-based): Measures how well marker confidence aligns with actual accuracy within the same domain (I-AvgECE) versus across different domains (C-AvgECE), compared against a standard numerical confidence baseline (NumECE).
  • Dispersion (CV-based): Measures how widely spread out or concentrated the confidence scores are within and across datasets (I-AvgCV, C-AvgCV).
  • Correlation: Tracks whether marker confidence moves proportionally with overall dataset accuracy (MAC) and whether the relative ranking of markers remains consistent (MRC).

Datasets

The framework was evaluated using seven diverse datasets requiring specialized knowledge across multiple domains:

  • Factual & Commonsense Knowledge: BoolQ, StrategyQA, CSQA
  • Mathematical Reasoning: GSM8K
  • Medical Reasoning: MedMCQA
  • Legal Reasoning: CaseHOLD
  • Mixed Knowledge: MMLU

Key Insights and Ideas

  • Good In-Domain, Terrible Out-of-Domain: LLMs are surprisingly well-calibrated when testing on the same data distribution they were evaluated on. However, when moving cross-domain, their verbal confidence calibration degrades significantly, often performing worse than simple numerical confidence prompts.
  • Shifting Definitions: The meaning of a word changes depending on the context. An LLM saying "quite certainly" might hit 90% accuracy on a straightforward factual dataset but drop to 50% accuracy on a complex legal or mathematical dataset.
Figure 2: Model’s marker confidence varies greatly across different datasets. We plot the heatmap of the marker confidence of GPT-4o and Qwen2.5-32B-Instruct across different datasets, illustrating that even the best models exhibit substantially different confidence levels in various contexts. The markers in the graph are randomly selected from the shared markers of all datasets.
  • Ranking Chaos: Models fail to maintain a logical hierarchy of words across domains. "Probably" might indicate higher accuracy than "seems" in one dataset, but flip completely in another.
Figure 3: The rankings of the model’s marker confidence fluctuates significantly across different datasets. We plot the scatter diagram for the marker confidence rankings of the best performing models, but still discovered that the rankings are extremely unstable. The markers in the graph are randomly selected from the shared markers across all datasets.
  • The "Uncertainty Compression" Problem: LLM marker confidences are highly concentrated. Models rarely use markers that correspond to very low accuracy (under 10%), showing a severe failure in expressing true, deep uncertainty through words.
  • Scale Helps: Larger, more powerful models (like GPT-4o and Qwen2.5-32B) exhibit significantly more stable marker rankings and lower cross-domain variance than smaller models.

Key Results

  • Calibration Breakdown: Across seven models, the In-domain Average Expected Calibration Error (I-AvgECE) was low at 8.17% (beating numerical confidence at 16.60%). However, the Cross-domain ECE (C-AvgECE) spiked to 17.73%, proving poor robustness.
  • Low Ranking Consistency: The average Marker Ranking Correlation (MRC) was a meager 21.34%, indicating that the relative order of confidence words is highly unstable across different tasks.
  • Model Performance Correlation: There is a strong negative correlation (-0.88) between a model's overall accuracy and its cross-domain variation (C-AvgCV), meaning smarter models use these markers more consistently.

Limitations and Future Work

  • Idealized Constraints: The study restricts models to closed-source QA tasks, English-only markers, and forces the output of exactly one marker per response, which strips away the natural complexity of human dialogue.
  • Lack of True Linguistic Alignment: The results show that LLMs mimic marker distributions from training data rather than possessing a robust, intrinsic understanding of verbal uncertainty.
  • Future Directions: Future work should explore how to align verbal confidence with actual model performance under distribution shifts and extend this evaluation to long-form, multi-turn communications.

Summary updated

Abstract

As large language models (LLMs) are increasingly used in high-stakes domains, accurately assessing their confidence is crucial. Humans typically express confidence through epistemic markers (e.g., "fairly confident") instead of numerical values. However, it remains unclear whether LLMs consistently use these markers to reflect their intrinsic confidence due to the difficulty of quantifying uncertainty associated with various markers. To address this gap, we first define marker confidence as the observed accuracy when a model employs an epistemic marker. We evaluate its stability across multiple question-answering datasets in both in-distribution and out-of-distribution settings for open-source and proprietary LLMs. Our results show that while markers generalize well within the same distribution, their confidence is inconsistent in out-of-distribution scenarios. These findings raise significant concerns about the reliability of epistemic markers for confidence estimation, underscoring the need for improved alignment between marker based confidence and actual model uncertainty. Our code is available at https://github.com/HKUST-KnowComp/MarConf.
BibTeX
@inproceedings{liu2025revisiting,
  author = {Jiayu Liu and Qing Zong and Weiqi Wang and Yangqiu Song},
  title = {{Revisiting Epistemic Markers in Confidence Estimation: Can Markers Accurately Reflect Large Language Models' Uncertainty?}},
  booktitle = {Annual Meeting of the Association for Computational Linguistics},
  year = {2025},
  doi = {10.18653/v1/2025.acl-short.18},
}
Figure