Using LLM Artificial Intelligence Systems as Complex SQL Programming Assistants

Abstract

Learning database programming such as SQL programming is a challenging task when the queries become more complex. SQL is a declarative language based on relational calculus which describes the definition of the query results instead of describing the procedure or steps used to obtain the query result. Tutorial sessions using tutorial assistances are generally required to support the learning of advanced part of the language. Recently generative AI systems demonstrated question answering capabilities including programming codes generation. This paper verifies the SQL code generating capabilities of four generative AI systems: Bing, Bard, ChatGPT, and Copilot and their suitability as SQL programming assistants.