Are LLMs Overkill for Databases?: A Study on the Finiteness of SQL
Yue Li, David Mimno, Unso Eun Seo Jo · 2026
Read Are LLMs Overkill for Databases?: A Study on the Finiteness of SQL by Yue Li, David Mimno, Unso Eun Seo Jo (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-25568.txt.
Description
Li, Mimno, and Jo ask whether giant language models are overkill for turning questions into SQL. They sampled 376 databases and measured the queries people actually need. Query complexity stays bounded in practice, and more tables do not mean harder queries. In template form, queries follow a steep power law. Thirteen percent of template types cover seventy percent of cases. That formulaic shape suggests templates could answer most needs more cheaply and with an audit trail. Language models may be solving a narrower problem than assumed.
Catalog record
| CREATOR | Yue Li, David Mimno, Unso Eun Seo Jo |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22748 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.25568