Developers argue AI coding quality depends on quality management systems, not AI capability

According to a Substack post by Iouri Khramtsov, the core issue when AI-assisted code lowers quality is inadequate quality management practices, not inherent AI limitations. Khramtsov describes a layered defensive approach combining spec-driven development, review processes, and testing that maintains or reduces bug rates while increasing output 2-2x. The argument frames AI coding productivity as compatible with quality when paired with structured governance rather than blind PR merging.