1. The Hidden Burden of AI Code
Generating pull requests with AI avoids effort for the writer while placing the entire weight of review and consumption on the reviewer.
Hear it at 0:00 →Engineering video
1:09 · 11 October 2026
The Hidden Cost of Reviewing AI Code
Key concepts in this video:
Transcript
And your coworkers will drown you with the PR that is fully written with AI. And the thing with AI is that it puts the weight on whoever will consume that. So, if you write a PR with AI, you don't pay the price. And I want to say you didn't even write a PR. You generated it. Okay, you just instructed the the AI to generate it. And the weight is on whoever is reviewing it. And we see a lot of companies, for example, they put the weight of product development, which is usually Q&A, on the users cuz they release crappy software. Then you find a bug, you report it, and they're like, "Oh, we fixed it the same day." I'm
like, "Yeah, this is not releasable yet. You're just pushing it out there, and now I'm doing the testing for you." What will happen in the long run is that products lose quality. It's easier to build competitors. And companies that are smart about how they use AI, they can actually use it but still have people building the decisions, building quality in the product, will grow. So, you see a lot of companies that are at the leadership level, they're not so AI-built. A good example would be Linear, for example, where their CTO has a really good grasp on AI. And they always talk about this human-first
approach, where they're like, "Engineers are using AI, and they are doing more with AI, but it's an engineering-centric approach to adopting AI, not And if you don't use it, we're going to lay you off."