Ahhhhhhhhhh hahahahahahahahahahaha. No. We have all seen how bad these LLMs are at coding, let's not have it review it's own failed coding attempt, too.
The End of Code Review: Coding Agents Supersede Human Inspection
11–20 of 22 posts
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#12Here is my take:
- I disagree with the contributions
- I suspect most of the paper was written / edited by "generative AI"
It is a shame that "researchers" have started using generative AI to such a large degree as it now masks the voice of the person. Generative AI tends to claim things that are not true and tends to use words that are unsuitable. This text leaves a bad taste in my mouth.
(Edits: formatting)
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#13Here is my take: - I disagree with the contributions - I suspect most of the paper was written / edited by "generative AI" It is a shame that "researchers" have started using generative AI to such a large degree as it now masks the voice of the person. Generative AI tends to claim things that are not true and tends to use words that are unsuitable. This text leaves a bad taste in my mouth. (Edits: formatting)
In a way, this discussion here is proof as to why human code review is necessary.
- The paper is garbage, and a human review process would reject it.
__As is the case in this discussion__. The sentiment of the discussion section (at the time of writing) seems to be in favor of rejecting this paper. (Of course, to make it a proper experiment, one would need to also give the paper to a "generative AI" reviewer to see if it would reject it or not, but I cannot bother.)
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#14Here is my take: - I disagree with the contributions - I suspect most of the paper was written / edited by "generative AI" It is a shame that "researchers" have started using generative AI to such a large degree as it now masks the voice of the person. Generative AI tends to claim things that are not true and tends to use words that are unsuitable. This text leaves a bad taste in my mouth. (Edits: formatting)
It seems they didn’t even do any research. This is not science, it’s just an opinion.
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#15Ironically, the number of sentences in bold tells me that an AI wrote this with no human review.
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#16And here I am huffing and puffing because the latest models keep adding nonsense to my codebase.
Have you seen the decisions LLMs make? They write code like the worst developers I know. They're lazy, short-sighted, and impossible to teach.
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#17It took 6 turns for opus 4.8 to attempt to make an illegal move in chess against me.
Yes you can setup guardrails and validations and other things, but as long its primary brain is demonstrably full of holes, I feel obligation to at very least be always be capable to take over the code. To me this means it should stay human-maintainable, and that's for humans to decide what is and isn't, one review at a time.
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#18I do think there's some real limits to code review as we do it and that we need to figure out what is next, to some degree. That's just my belief here. Orgs keeping to the past are going to be kind of miserable; leads already had a ton of responsibility for understanding it all, and it feels like that just isn't feasible to keep on with.
That's a pretty bike take, but I think a bit less than what this paper is saying.
A more mild sub-argument I really enjoyed recently was just that we have basically an obligation to have AI code review:
> Nuclear opinion: I don’t care if you use AI to write code,¹ but making a human review a change before an LLM is disrespectful.
> Why am I catching your bugs if a computer could have done it? It’s like asking to review code that doesn’t compile or pass CI.
https://bsky.app/profile/filippo.abyssdomain.expert/post/3mo...
This makes me also really appreciate a thread that shows up recently, where folks were talking about their code/repo forges. Using tickets/PRs as rendezvous for where people and agents come together had very high appeal to me. Being able to surface discussion in a shared context seems crucial.
https://btao.org/posts/2026-05-09-the-forge-we-deserve/ https://news.ycombinator.com/item?id=48582147
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#19And here I am huffing and puffing because the latest models keep adding nonsense to my codebase. Have you seen the decisions LLMs make? They write code like the worst developers I know. They're lazy, short-sighted, and impossible to teach.
[dead]
Re: The End of Code Review: Coding Agents Supersede Human Inspection
#20I strongly disagree with "Claim 1: Every goal of code review can be served by agents at lower cost and higher throughput." They define the goals and include this one:
> "Knowledge transfer. Agents can actively generate on-demand explanations, architectural summaries, and updated documentation at merge time, which is a more reliable and scalable mechanism for propagating knowledge than the incidental commentary of a busy colleague.
Knowledge transfer is much more than what they claim. An critical goal of code review is making sure humans do deeply understand overall what software actually does. Code is far more precise and detailed than any "on-demand explanations, architectural summaries, and updated documentation."