# Disentangling Ambiguity from Instability in Large Language Models: A Clinical Text-to-SQL Case Study

arXiv: https://arxiv.org/abs/2602.12015 · HTML: https://mlsanity.com/paper/2602.12015/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 40 references that are in the map.
- First posted on arXiv 12 Feb 2026.

## Builds on (most cited first, 21 of 21)

- [Convex optimization](https://mlsanity.com/paper/2106.01946/) (2021)
- [What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?](https://mlsanity.com/paper/1703.04977/) (2017)
- [Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Task](https://mlsanity.com/paper/1809.08887/) (2018)
- [Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methods](https://mlsanity.com/paper/1910.09457/) (2019)
- [AmbigQA: Answering Ambiguous Open-domain Questions](https://mlsanity.com/paper/2004.10645/) (2020)
- [Kernel Language Entropy: Fine-grained Uncertainty Quantification for LLMs from Semantic Similarities](https://mlsanity.com/paper/2405.20003/) (2024)
- [SituatedQA: Incorporating Extra-Linguistic Contexts into QA](https://mlsanity.com/paper/2109.06157/) (2021)
- [Recent Advances in Text-to-SQL: A Survey of What We Have and What We Expect](https://mlsanity.com/paper/2208.10099/) (2022)
- [Distinguishing the Knowable from the Unknowable with Language Models](https://mlsanity.com/paper/2402.03563/) (2024)
- [AMBROSIA: A Benchmark for Parsing Ambiguous Questions into Database Queries](https://mlsanity.com/paper/2406.19073/) (2024)
- [Semantic Volume: Quantifying and Detecting both External and Internal Uncertainty in LLMs](https://mlsanity.com/paper/2502.21239/) (2025)
- [Disambiguate First, Parse Later: Generating Interpretations for Ambiguity Resolution in Semantic Parsing](https://mlsanity.com/paper/2502.18448/) (2025)
- [A Bias-Variance-Covariance Decomposition of Kernel Scores for Generative Models](https://mlsanity.com/paper/2310.05833/) (2023)
- [Retrieval augmented text-to-SQL generation for epidemiological question answering using electronic health records](https://mlsanity.com/paper/2403.09226/) (2024)
- [ProbGate at EHRSQL 2024: Enhancing SQL Query Generation Accuracy through Probabilistic Threshold Filtering and Error Handling](https://mlsanity.com/paper/2404.16659/) (2024)
- [LG AI Research & KAIST at EHRSQL 2024: Self-Training Large Language Models with Pseudo-Labeled Unanswerable Questions for a Reliable Text-to-SQL System on EHRs](https://mlsanity.com/paper/2405.11162/) (2024)
- [Inv-Entropy: A Fully Probabilistic Framework for Uncertainty Quantification in Language Models](https://mlsanity.com/paper/2506.09684/) (2025)
- [Generating patient cohorts from electronic health records using two-step retrieval-augmented text-to-SQL generation](https://mlsanity.com/paper/2502.21107/) (2025)
- [BiomedSQL: Text-to-SQL for Scientific Reasoning on Biomedical Knowledge Bases](https://mlsanity.com/paper/2505.20321/) (2025)
- [Semantic uncertainty in advanced decoding methods for LLM generation](https://mlsanity.com/paper/2506.17296/) (2025)
- [A Survey on LLM-Assisted Clinical Trial Recruitment](https://mlsanity.com/paper/2506.15301/) (2025)

## Built on by (most cited first, 3 of 3)

- [Ascent: An Agentic System over the Model Context Protocol for Real-World Clinical Data Analysis](https://mlsanity.com/paper/2609.24620/) (2026)
- [Certified Against Which Oracle? Execution Labels Set the Reported Risk of Conformal Abstention for Text-to-SQL](https://mlsanity.com/paper/2609.25938/) (2026)
- [What Is a Clarification? Estimating Model-Dependent Input Ambiguity](https://mlsanity.com/paper/2610.20052/) (2026)

## Abstract (arXiv)

Deploying large language models for clinical Text-to-SQL requires distinguishing two qualitatively different causes of output diversity: (i) input ambiguity that should trigger clarification, and (ii) model instability that should trigger human review. We propose CLUES, a framework that models Text-to-SQL as a two-stage process (interpretations --> answers) and decomposes semantic uncertainty into an ambiguity score and an instability score. The instability score is computed via the Schur complement of a bipartite semantic graph matrix. Across AmbigQA/SituatedQA (gold interpretations) and a clinical Text-to-SQL benchmark (known interpretations), CLUES improves failure prediction over state-of-the-art Kernel Language Entropy. In deployment settings, it remains competitive while providing a diagnostic decomposition unavailable from a single score. The resulting uncertainty regimes map to targeted interventions - query refinement for ambiguity, model improvement for instability. The high-ambiguity/high-instability regime contains 51% of errors while covering 25% of queries, enabling efficient triage.