# The Stochastic Shift: A New Evaluation Paradigm for Text-to-SQL with AI Operators

arXiv: https://arxiv.org/abs/2609.21133 · HTML: https://mlsanity.com/paper/2609.21133/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 17 references that are in the map.
- First posted on arXiv 17 Sep 2026.

## Builds on (most cited first, 11 of 11)

- [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)
- [DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines](https://mlsanity.com/paper/2310.03714/) (2023)
- [Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs](https://mlsanity.com/paper/2305.03111/) (2023)
- [Semantic Evaluation for Text-to-SQL with Distilled Test Suites](https://mlsanity.com/paper/2010.02840/) (2020)
- [Natural SQL: Making SQL Easier to Infer from Natural Language Specifications](https://mlsanity.com/paper/2109.05153/) (2021)
- [SQL-PaLM: Improved Large Language Model Adaptation for Text-to-SQL (extended)](https://mlsanity.com/paper/2306.00739/) (2023)
- [Semantic Operators: A Declarative Model for Rich, AI-based Data Processing](https://mlsanity.com/paper/2407.11418/) (2024)
- [Evaluating Cross-Domain Text-to-SQL Models and Benchmarks](https://mlsanity.com/paper/2310.18538/) (2023)
- [Grading Scale Impact on LLM-as-a-Judge: Human-LLM Alignment Is Highest on 0-5 Grading Scale](https://mlsanity.com/paper/2601.03444/) (2026)
- [TrustSQL: Benchmarking Text-to-SQL Reliability with Penalty-Based Scoring](https://mlsanity.com/paper/2403.15879/) (2024)
- [Cortex AISQL: A Production SQL Engine for Unstructured Data](https://mlsanity.com/paper/2511.07663/) (2025)

## Built on by (most cited first, 1 of 1)

- [Same Query, Different Answer: Run-to-Run Variance in Semantic Query Processing](https://mlsanity.com/paper/2610.13823/) (2026)

## Abstract (arXiv)

SQL has been augmented with AI operators, enabling modern data analytics platforms to derive insights from both structured and unstructured data. We observe that while current Text-to-SQL systems can successfully generate these AI-augmented queries, reliably evaluating their correctness remains a critical open challenge. Current metrics, which rely on exact query results and deterministic execution, systematically fail against the flexible, non-deterministic outputs of AI operators. In this paper, we formalize these unique evaluation failure modes and introduce a Multilayered Evaluation Framework that decouples deterministic database logic from flexible AI semantics. We test our approach across both industry (BigQuery) and academic (ThalamusDB) systems. We demonstrate that traditional Execution Accuracy severely penalizes valid queries, achieving as low as a 25% detection rate for correct translations. Furthermore, even a state-of-the-art LLM-based autorater falsely rejects 32% of accurate queries due to the complexity of judging both relational and AI components simultaneously. By validating the standard relational logic and the AI operations separately, our framework achieves state-of-the-art overall accuracy across both platforms (up to 97.2%), proposing a reliable standard for benchmarking AI-powered SQL generators.