frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

The Missed Reality: Code Review Wasn't Built for the AI Era

2•axsdrizz•15h ago
The Bottleneck Isn't Writing Code Anymore. It's Trusting It.

The speed at which AI generates code has officially surpassed our capacity to review it. For decades, the bottleneck in software engineering was writing the code. Today, the bottleneck has shifted entirely to trusting that the code can run effectively in production and actually solve the business problem.

But let's be honest: humans never truly scaled code review anyway.

Throughout my career, I have observed developers approving PRs or MRs that contain critical bugs. When they do engage in the review process, they often focus on stylistic issues. Why does this happen? Because reviewers are overloaded with their own backlogs to manage, and often choose the path of least resistance by saying, "Looks good to me."

Now, we are trying to solve this by throwing AI at the problem!

The Missed Reality: Code Review Wasn't Built for the AI Era

Code review - whether conducted by a human or an LLM - was not designed for this era of autonomous coding. Code review does not actually verify whether the code accomplishes what was prompted, nor do they confirm that the code aligns with its intended specification or that an agent’s "task completed" claim is deterministically true. Even a well-conducted review adhering to the 5-axes best practices (correctness, architecture, security, readability, and performance) falls short. Rarely does a code review verify that the original acceptance criteria were met or that the AI agent didn't deviate from the scope just to finish the PR. LLMs notoriously defer hard, multi-step, or dependent tasks, which means that the intent itself has diverged. Furthermore, reviewers rarely verify that test coverage was maintained or check if the AI agent quietly weakened the tests along the way to secure a passing PR.

This leads to a critical question: Even if a human reviewer or an LLM could effectively manage all these new checks, can they realistically and deterministically do so given the massively increased speed of AI code generation?

The solution? We have to move from opinion-based review to evidence-based verification.

This means building guardrails (what I call the claim checker) that bind the original intent to deterministic evidence: ACs (intents) checks, the builds, the tests, the five-axis code review, the scans, the blast radius of the code change, the docs coverage, requirements are complete and don't contradict each other, and the monitoring coverage in the code.

AI can advise. But only evidence should decide whether to merge or not!

It’s a shift we probably should have made years ago. But with agent loops writing code at machine speed, it is now the only way forward.

Comments

rishabhpoddar•14h ago
I was actually working on something like a "claim checker" recently. Inspired by stuff like Codeql. But I realised that all the really valuable business facing claims can't be checked just by static code analysis, and need to be done via integration and unit tests only. Given that claude code can pretty much write all the runtime tests fairly easily and quickly, I don't see how anything else can be better.

Ask HN: Who gets credits on big math questions solved by LLMs?

5•silentmafia•1h ago•1 comments

How do you guys keep up with AI news?

4•anirudhc2005•3h ago•4 comments

Learn by Building

7•csacademy•7h ago•2 comments

Ask HN: What Are You Working On? (July 2026)

282•david927•4d ago•1092 comments

X permanently banned my 15-year-old account for sharing my own open-source lib

11•rifmj•4h ago•4 comments

Ask HN: Is it just me, or is software buggier across the board?

124•kadhirvelm•1d ago•105 comments

Why people chasing after useless token saving plugins and ignoring real solution

2•yohji1984•5h ago•0 comments

Ask HN: How do you manage too many open tabs?

7•NarcisMirandes•9h ago•22 comments

Ask HN: Are we there yet? Can I create a human analogue using AI?

2•JoeAltmaier•6h ago•3 comments

Ask HN: As LLMs progress, how do you stay sharp and productive?

7•spiresofagartha•10h ago•7 comments

Ask HN: Best meeting transcription daemon for macOS?

4•tomelliot•7h ago•3 comments

Ask HN: Assistive Tech Ideas for Dementia

3•thanksd•9h ago•1 comments

Ask HN: Add flag for AI-generated articles

1084•levkk•3d ago•457 comments

Thinking: Alternate Distribution for Filmmakers

3•phaedrus044•15h ago•0 comments

Ask HN: Are AWS Certs worth getting for experienced Devs?

10•Gold-Ask-lx•1d ago•16 comments

Help me be accountable with my time

4•atomiomi•11h ago•10 comments

Ask HN: How about a "Merge Stories" feature?

2•RetroTechie•11h ago•2 comments

Ask HN: What are you using for authn/authz and payments?

3•jftuga•12h ago•4 comments

Ask HN: Are there any startups looking for a developer on contract base?

2•need_a_work23•13h ago•1 comments

OTPme Testers Wanted

4•otpme•13h ago•0 comments

On-device AI that respects your privacy

2•fujidevv•13h ago•0 comments

Ask HN: How do you handle OSS project with private artifacts?

2•mrsalty•14h ago•1 comments

The Missed Reality: Code Review Wasn't Built for the AI Era

2•axsdrizz•15h ago•1 comments

Ask HN: Should I do a CS masters at Cambridge or start as a new grad at Amazon?

28•sspehr•3d ago•93 comments

Ask HN: Best way to buy second hand compute (in a future AI crash)

4•lifeisstillgood•15h ago•1 comments

I built a tool to get comments in CSV format from YouTube videos

2•cristyg0101•17h ago•2 comments

Laptop touchpad stopped registering clicks? try this

5•LandOfMightDev•18h ago•1 comments

I'd prefer to avoid AI comments on Reddit and HN

5•alance•8h ago•1 comments

You've reached the end!