NERD is a new programming language designed for an era where AI generates most code. As of now, 40% of code comes from large language models (LLMs), a trend that’s expected to grow. The creators of NERD argue that traditional programming languages are not optimized for AI-generated code. They propose a shift away from human-readable formats towards a machine-optimized structure that requires less effort from humans. This change aims to simplify the coding process and reduce the number of tokens needed, claiming a reduction of 50-70% compared to languages like TypeScript while maintaining the same functionality. The design philosophy behind NERD focuses on making code dense and terse, using plain English words rather than complex symbols. For instance, a simple function to add two numbers can be expressed with significantly fewer tokens in NERD than in TypeScript. The workflow emphasizes that humans become stakeholders in the coding process rather than authors, where LLMs write the code and humans provide oversight. This approach challenges conventional notions of debugging and compliance, suggesting that auditing can be done through translated views of the code instead of requiring human-readable syntax. Critics raise concerns about debugging and readability, but the creators argue that these issues can be addressed at higher abstraction levels. They believe that as AI takes over more coding tasks, the need for human-written code will diminish. NERD is positioned as a prototype, with the potential to redefine how software is developed in the future, though the authors acknowledge the uncertainty of its acceptance and effectiveness.