You are absolutely right? – LLM workflows and thoughts about the future (cpojer.net)

In a recent discussion, Christoph Nakazawa highlighted the seismic changes anticipated in software engineering due to advancements in AI, specifically in the use of Large Language Models (LLMs) like Codex and ChatGPT. Previously, LLMs struggled with complex coding tasks, but Nakazawa notes that they have now evolved to significantly aid in the development of substantial applications. His experience reflects a surge in AI-assisted project work, with AI integrated into nearly all aspects of his recent projects, such as "fate," a modern data library for React, and "Relang.dev," a service for auto-translating JavaScript apps. Nakazawa emphasizes the importance of using multiple solution outputs to enhance coding efficiency and reduce errors, as generating multiple responses often leads to deeper insights and faster resolutions for complex issues. He also shares his preference for maintaining control over the coding process, opting to heavily edit AI-generated code rather than relying on it completely. This striking balance between human oversight and AI assistance showcases the potential for LLMs to augment rather than replace traditional coding skills, underscoring their growing significance in the AI/ML community as tools that streamline development and improve productivity while still necessitating meticulous human craftsmanship for quality control.