Are AI SQL Yet? (www.viblo.se)

Recent advancements in AI, particularly the release of models like Opus 4.6 and GPT Codex 5.3, have led to a fresh evaluation of AI's capabilities in handling complex SQL database tasks. Despite previous disappointments, the author tested these new models for support in renaming a column in a critical temporal table, which monitors account balances in a fintech application. The challenge lay in the need to turn off and on the temporal feature while ensuring minimal downtime and avoiding locking issues on a production database with significant but not massive user activity. The experiment revealed that although only one AI provided a robust response, it highlighted crucial elements such as the need for a Schema Modification Lock during the renaming process. This lock can block all access to the table, potentially leading to a "convoy effect" if a long-running query is executing. The optimal solution included adjusting locking strategies and using parameters like `DATA_CONSISTENCY_CHECK = OFF` to reduce unnecessary downtime, demonstrating how current AI models can offer valuable insights for operational efficiency in database management. This evolution in AI capabilities is significant for the AI/ML community, affirming that recent advancements can indeed enhance practical applications in complex technical scenarios.