Seems pretty spot on, based on my extremely limited experience thus far with agentic development. If I don't narrowly scope & carefully review the changes the agent makes, the potential for it making changes that cascade into bigger problems explodes. At which point, I spend a lot of subsequent time and effort identifying and fixing those problems.

Reads like LLM text that says little, and talks surface level. Wasn't able to read through it.

> But the shift is real. When Karpathy admits he barely writes code directly anymore, when the Claude Code team ships 20+ PRs daily with 100% AI-written code, we’re past the point of dismissing this as hype.

Oh a company 100% invested in AI hype says they have 100% AI-written code PRs. Water is still wet. CC Team is least trustable team when it comes to judging how good AI is.

Also, to Pro-AI people, if you really believe that models are going to become much better from now, why shouldnt we just wait for those models rather than building tech debt at a scale unseen before 2025?