A lot of AI-coding enthusiasts are making claims along the lines of “AI coding tools are like compilers; you supply intent, they translate it into code, and who ever looked at the machine-code output?” Unfortunately, there is a bit of a gap between hope and reality; traditional compilers were always deterministic, and are (after decades of development and bug-fixing) pretty much bug-free. AI coding tools are neither, and no amount of “agentic loops” will solve that.

Not surprisingly, people with long enough AI exposure are slowly discovering that gap, as Cal Newport (well worth following, even if you disagree with any of his takes) described in his On AI Coding and Its Discontents blog post.