Specification; communication; computation – no, programming isn't dead (twey.io)

A recent article critiques the prevailing belief that programming is becoming obsolete due to advancements in large language models (LLMs), which allow individuals to 'code' using natural language. While tech influencers celebrate AI’s potential to simplify coding, studies indicate that the output of real, complex software hasn’t significantly improved, revealing that the perception of increased productivity is misleading and leads to developer burnout. The fundamental issue is that LLMs excel at language-based tasks but lack the ability to perform actual programming, which requires intricate decision-making and problem-solving beyond mere syntax translation. The author argues that programming languages themselves are designed to aid human comprehension, not just to serve machine demands. Simplifying programming to a natural language interface overlooks the complexity and nuance required in decision-making processes involved in software development. Ultimately, while LLMs offer value, they cannot replace the critical thinking and expertise that professional programmers bring to software development, suggesting that programming in its true sense is far from dead and remains essential for developing sophisticated, meaningful applications in a technology-driven world.