The Four Horsemen of Agentic Coding (distantprovince.substack.com)

A recent discussion highlights the complex challenges posed by agentic coding, which relies on large language models (LLMs) to generate code. The term "agentic coding" refers to the growing reliance on AI to automate coding tasks, which, while enhancing productivity, brings about significant drawbacks: poor code quality ("slop"), alienation from the coding process, deskilling of engineers, and deterioration of team dynamics. The prevalence of sloppy and sometimes unintelligible code raises concerns about the ability to maintain a cohesive and collaborative coding environment. These developments have significant implications for the AI/ML community. The detachment from the coding process leads to a lack of ownership and pride in work, with individuals relying heavily on AI instead of honing their skills. As teams increasingly communicate through agents rather than direct interaction, the rich interpersonal relationships that traditionally characterize software development are weakened. Given these challenges, experts express concern about the long-term consequences on skill development and team cohesion, emphasizing the need for ongoing dialogue on the balance between AI utility and the human aspects of coding.