Research map / Large language models / Language, safety and interpretability
“very likely” Means “uncertain”? How LLMs Diverge from Humans in Linguistic Uncertainty Quantification
Jinhao Duan, Zicheng Liu, Zijie Liu, Kaidi Xu, Tianlong Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| learn LLM-specific marker scores (VOCAL) instead of human-sourced UM-Lookup | VOCAL · GSM8K, GPT-4o | AUROC: 0.41 -> around 0.6 | Improved |
| use VOCAL instead of single-sample G-NLL and PPL | VOCAL · GSM-Hard, DeepSeek-V3.1 | AUROC: 0.567 -> 0.715 | Improved |
| VOCAL vs PPL on Trivia QA with GPT-4o | VOCAL · Trivia QA, GPT-4o | AUROC: 0.575 vs 0.573 | No clear effect |
| increase training samples from 100 to 500 (ablation) | VOCAL · UQ performance | AUROC: 0.52 -> 0.60 | Improved |
| vary graph Laplacian smoothing strength gamma away from 0.005 (ablation) | VOCAL · AUROC ablation | AUROC | Hurt |
| apply isotonic regression post-hoc to VOCAL | VOCAL · GPT-4o, avg of GSM-Hard/GSM8K/MedQA | ECE: 0.1314 -> 0.0281 | Improved |
| apply temperature scaling post-hoc to VOCAL | VOCAL · GPT-4o, avg of GSM-Hard/GSM8K/MedQA | ECE: 0.1314 -> 0.0360 | Improved |
| use verbal uncertainty prompting instead of standard CoT | GPT-4o to Llama-3.2-3B · GSM8K | accuracy | No clear effect |
| transfer learned lookup across LLMs | VOCAL · SciQ cross-LLM | AUROC: 0.60-0.72 | Mixed |
Closest papers by meaning
- Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs (ICLR 2024 · similarity 0.90)
- Calibrating Expressions of Certainty (ICLR 2025 · similarity 0.89)
- Position: Uncertainty Quantification Needs Reassessment for Large Language Model Agents (ICML 2025 · similarity 0.89)
- To Believe or Not to Believe Your LLM: Iterative Prompting for Estimating Epistemic Uncertainty (NeurIPS 2024 · similarity 0.88)
- Large Language Models Must Be Taught to Know What They Don’t Know (NeurIPS 2024 · similarity 0.87)
- Benchmarking LLMs via Uncertainty Quantification (NeurIPS 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
