Human coders are still better than LLMs

The author argues that despite the utility of large language models (LLMs) like Gemini in development tasks, human coders still outperform them when it comes to creative problem-solving. The post illustrates this with a personal story of debugging and optimizing code for Redis vector sets. While LLMs can suggest solutions, the author emphasizes the complexity and originality of human thought that enables innovative solutions beyond the current capabilities of AI.