AI from the perspective of a software developer, year-end 2025 (www.themotte.org)

As we approach the end of 2025, software developers are experiencing a significant transformation in their coding practices due to advancements in AI technologies, primarily driven by large language models (LLMs). The article reflects on how developers have progressively shifted from skepticism to embracing AI, with 2025 marking a concrete change in how coding is approached. While many developers are utilizing AI tools to increase their productivity by 20-50%, there remains a cautionary note about potential future risks to employment as AI capabilities continue to improve. Importantly, the author emphasizes that while AI can facilitate tasks like translating code and managing front-end development, human oversight is still crucial for context and troubleshooting. In 2025, the improvements in AI were notable, with advancements in models such as Gemini Pro 2.5 and ChatGPT o3 achieving impressive results in coding tasks. However, the year witnessed mixed outcomes; while there were improvements in AI tools for code editing and contextual understanding, many developers remain skeptical about the complete automation of their roles. The text highlights the need for further advancements in AI agents that can handle more complex general programming tasks. Despite the hype surrounding AI's potential, developers are reminded that practical usage still requires human intervention, with an ongoing need to balance AI’s efficiency against its limitations in decision-making and creative processes.