MIT researchers have introduced a new AI tool that can help database users perform complicated statistical analyses of tabular data without knowing what is going on behind the scenes.

GenSQL, a generative AI system for databases, could help users make predictions, detect anomalies, guess missing values, fix errors, or generate synthetic data with just a few keystrokes. It can automatically integrate a tabular dataset and a generative probabilistic AI model, which can account for uncertainty and adjust their decision-making based on new data. It can also produce and analyze synthetic data that mimic the real data in a database. This could be especially useful when sensitive data, such as patient health records, cannot be shared or when real data are sparse.

GenSQL is built on top of SQL, a programming language for database creation and manipulation that was introduced in the late 1970s and is used by millions of developers worldwide.

According to Vikash Mansinghka ’05, MEng ’09, PhD ’09, senior author of a paper introducing GenSQL and a principal research scientist and leader of the Probabilistic Computing Project in the MIT Department of Brain and Cognitive Sciences, historically, SQL taught the business world what a computer could do. They didn’t have to write custom programs; they just had to ask questions of a database in a high-level language. We think that, when we move from just querying data to asking questions of models and data, we are going to need an analogous language that teaches people the coherent questions you can ask a computer that has a probabilistic model of the data.

Faster and accurate

When the researchers compared GenSQL to popular, AI-based approaches for data analysis, they found that it was not only faster but also produced more accurate results. Importantly, the probabilistic models used by GenSQL are explainable, so users can read and edit them. SQL, which stands for structured query language, is a programming language for storing and manipulating information in a database. In SQL, people can ask questions about data using keywords, such as by summing, filtering, or grouping database records.

The researchers noticed that SQL didn’t provide an effective way to incorporate probabilistic AI models, but at the same time, approaches that use probabilistic models to make inferences didn’t support complex database queries. They built GenSQL to fill this gap, enabling someone to query both a dataset and a probabilistic model using a straightforward yet powerful formal programming language.

A GenSQL user uploads their data and probabilistic model, which the system automatically integrates. Then, she can run queries on data that also get input from the probabilistic model running behind the scenes. This not only enables more complex queries but can also provide more accurate answers.

To evaluate GenSQL, the researchers compared their system to popular baseline methods that use neural networks.GenSQL was between 1.7 and 6.8 times faster than these approaches, executing most queries in a few milliseconds while providing more accurate results.

MIT