Hacker News Reader

Reviewing agent-generated code is a pain point.

I think the traditional code review process is not sustainable with the new speed of delivery. Should we just execute automatic code review leveraging the usual PRD+specs with the addition of a detailed architectural document?

View on HN

I think we should let reviews go. The next in the list are tests and CI, they also slow down delivery, and should go too.

Then, we'll have the full speed and the consequences of that. I want to see it happening on a massive scale.

Human can't review AI generated code at this speed. It is impossible. We have already let the reviews go because the majority of engineers are approving changes without truly understanding it.

Yes, yes. Let's do that with testing and QA too. In massive scales. Humans are too slow. I want to see what happens.

Reply on news.ycombinator.com