As a Data Engineer and Analyst, I’ve been experimenting with ChatGPT and Claude to write complex SQL queries. While they’re great at understanding BigQuery syntax, I’ve hit a roadblock when it comes to crafting queries with intricate logic and filtering. No matter how well I prompt them, they either hallucinate or mess up complex window functions and logic.
I’m left wondering: am I bad at prompt-engineering, or do these AI tools have a fundamental limitation when it comes to complex SQL queries? I’ve tried to optimize my workflow by asking ChatGPT to generate a prompt for my requirements, then sending it to Claude to write the SQL query. But even with this approach, I’m not getting the results I need.
Maybe I’m expecting too much from these AI tools. After all, writing complex SQL queries requires a deep understanding of the data, the schema, and the relationships between tables. It’s possible that AI just isn’t ready to take on this task – at least, not yet.
So, what’s the solution? Should I focus on improving my prompt-engineering skills, or should I just write the queries myself? The time it takes to get ChatGPT or Claude to produce a decent query isn’t justified when I can do it myself. But then again, maybe I’m missing something. Maybe there’s a better way to work with these AI tools to generate complex SQL queries.
What do you think? Have you had similar experiences with ChatGPT or Claude? Share your thoughts in the comments!