--- title: Code review is broken and AI just made it impossible to... description: AI-generated code is causing a volume crisis in code review — Meta reports a 106% increase in lines of code per diff and 51% more diffs per developer. The... canonical: https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Code review is broken and AI just made it impossible to ignore | daily.dev og:description: AI-generated code is causing a volume crisis in code review — Meta reports a 106% increase in lines of code per diff and 51% more diffs per developer. The... og:url: https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy og:image: https://api.daily.dev/og/posts/PKM4HQlHY.png og:image:alt: Code review is broken and AI just made it impossible to ignore og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Code review is broken and AI just made it impossible to ignore **[Trends](https://daily.dev/sources/trends)** · 2 min read · 4 upvotes · 2 comments ## Summary AI-generated code is causing a volume crisis in code review — Meta reports a 106% increase in lines of code per diff and 51% more diffs per developer. The traditional pre-merge review gate is breaking down. Only 14% of review comments catch actual defects; the rest serves knowledge transfer and shared understanding. The proposed fix isn't to eliminate review but to redesign it: move review upstream to intent and acceptance criteria, codify invariants to catch recurring mistakes automatically, and use advisory agents during generation rather than after commit. Meta's RADAR system demonstrates the model — automating low-risk diffs and routing high-risk ones to humans — achieving a revert rate one-third of normal and a production incident rate one-fiftieth as high. The warning: teams that haven't fixed basics like PR size and change descriptions will just accelerate their dysfunction. ## Content The line that kicked off this whole fight: **@svpino** announced he's "officially done reading AI-generated code," hasn't looked at a diff in two weeks, and thinks the IDE is heading to the graveyard. He's not alone. **@ThePrimeagen** joked about merging 10 PRs a day from a cloud agent and wondering if that makes him "officially left behind." **@Hesamation** put the whiplash in perspective: nine months ago everyone insisted they coded better than AI. Now everyone's admitting they don't even review its output. The pushback was immediate and pointed. **@rauchg** laid out a blunt checklist: if you're not reading the code, you're probably a beginner, prototyping, have no real users, or are quietly taking on debt and risk. He's watched frontier models add a nonsensical 700ms delay and then admit to "cargo-culting" when questioned. **@willmcgugan** went further and disabled PRs entirely on Textual, calling it resume-padding slop: "Stop. Don't slop!" **@thdxr** reframed the whole debate as a red herring, arguing the real bar hasn't moved: if you own the software, you should be able to explain how it works from memory, whether you typed it or not. The data backs up both sides being a little right. Meta's diffs grew 106% in size and 51% in volume per developer, driven by agentic AI, while Stripe now ships 1,300 PRs a week. But the DORA report found every 25% jump in AI adoption correlates with a 7.2% drop in delivery stability. LinearB's 2026 benchmarks show agentic PRs merging at just 37% in weaker orgs versus 79% in stronger ones. And Dex Horthy's SlopCodeBench found unreviewed AI output had 11x more ## Questions this post answers ### Does higher AI coding tool adoption actually hurt software delivery stability? Yes, according to DORA research cited in the debate, every 25% increase in AI adoption correlates with a 7.2% drop in delivery stability. This finding is used to argue that merging AI-generated pull requests without review introduces meaningful risk, even as teams like Meta and Stripe dramatically scale up AI-driven code output. _Engineers weighing AI-assisted workflows against stability risks can follow this debate as it develops on daily.dev._ ### What percentage of AI-generated pull requests get merged without human review? LinearB's 2026 benchmarks found agentic pull requests merge at just 37% in weaker-performing organizations versus 79% in stronger ones, suggesting review discipline varies widely by team maturity. Separately, Dex Horthy's SlopCodeBench found unreviewed AI output contained 11 times more issues than reviewed code. _Teams setting merge policies for agentic PRs can track this evolving benchmark data on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 1 comments across x (as of 2026-09-13). **TL;DR:** There isn't enough discussion content provided to characterize a community reaction beyond the fact that the topic is contentious by nature. **Sentiment:** 10% positive · 20% mixed · 70% skeptical **By community** - x (mixed): No substantive replies were provided to gauge sentiment. **Source threads** - [x](https://x.com/QodoAI/status/2088271365023412281) · 0 points · 1 comments ## Community discussion Top comments from developers on daily.dev. **@agustinbarrientos** · 1 upvotes > For me, one human still needs to explain the failure modes before a risky diff merges. Acceptance criteria rarely cover what the team failed to imagine. **@parsam** · 0 upvotes > the part i keep coming back to is that reviewing the diff was never the same thing as reviewing the change. > > AI makes the diff cheaper to produce, so obviously we're going to have more of them. the harder question is what those changes actually affect. > > a 500 line PR isn't necessarily more concerning than a 5 line change to something shared. > > that's where i'd rather have a production reliability signal alongside the usual review, something that looks at dependencies, fragility, past incidents and blast radius instead of just the code itself. ## Similar posts on daily.dev - [Why AI Code Reviews Are Slowing Down PRs](https://daily.dev/posts/why-ai-code-reviews-are-slowing-down-prs-yubjqxjqr) · Aviator Blog · 0 upvotes · 1 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#productivity](https://daily.dev/tags/productivity), [#code-review](https://daily.dev/tags/code-review), [#technical-debt](https://daily.dev/tags/technical-debt) [View this post on daily.dev](https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Code review is broken and AI just made it impossible to ignore","url":"https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy"},"datePublished":"2026-08-08T12:45:23.771Z","dateModified":"2026-09-13T19:44:55.166Z","description":"AI-generated code is causing a volume crisis in code review — Meta reports a 106% increase in lines of code per diff and 51% more diffs per developer. The...","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":2,"discussionUrl":"https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":4},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":2}],"keywords":"ai-agents,productivity,code-review,technical-debt","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Code review is broken and AI just made it impossible to ignore"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy","comment":[{"@type":"Comment","text":"For me, one human still needs to explain the failure modes before a risky diff merges. Acceptance criteria rarely cover what the team failed to imagine.","datePublished":"2026-08-08T20:28:56.343Z","url":"https://daily.dev/posts/PKM4HQlHY#c-dQ8nl9aG7","author":{"@type":"Person","name":"Agustin Barrientos","url":"https://daily.dev/agustinbarrientos","image":"https://media.daily.dev/image/upload/s--5ayxQnqn--/f_auto/v1788281802/avatars/avatar_wQYYVe5Tbj0NJ7C7qPoa8?_a=BAMAMicg0"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}},{"@type":"Comment","text":"the part i keep coming back to is that reviewing the diff was never the same thing as reviewing the change.\nAI makes the diff cheaper to produce, so obviously we’re going to have more of them. the harder question is what those changes actually affect.\na 500 line PR isn’t necessarily more concerning than a 5 line change to something shared.\nthat’s where i’d rather have a production reliability signal alongside the usual review, something that looks at dependencies, fragility, past incidents and blast radius instead of just the code itself.","datePublished":"2026-09-30T16:47:13.354Z","url":"https://daily.dev/posts/PKM4HQlHY#c-95qooItwf","author":{"@type":"Person","name":"Parsa","url":"https://daily.dev/parsam","image":"https://lh3.googleusercontent.com/a/ACg8ocI7cLcjVdmLqIn3k8yJcwf5xcUYqgUd6d2dh6oiY59utH1geA=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/code-review-is-broken-and-ai-just-made-it-impossible-to-ignore-pkm4hqlhy#faq","mainEntity":[{"@type":"Question","name":"Does higher AI coding tool adoption actually hurt software delivery stability?","acceptedAnswer":{"@type":"Answer","text":"Yes, according to DORA research cited in the debate, every 25% increase in AI adoption correlates with a 7.2% drop in delivery stability. This finding is used to argue that merging AI-generated pull requests without review introduces meaningful risk, even as teams like Meta and Stripe dramatically scale up AI-driven code output. Engineers weighing AI-assisted workflows against stability risks can follow this debate as it develops on daily.dev."}},{"@type":"Question","name":"What percentage of AI-generated pull requests get merged without human review?","acceptedAnswer":{"@type":"Answer","text":"LinearB's 2026 benchmarks found agentic pull requests merge at just 37% in weaker-performing organizations versus 79% in stronger ones, suggesting review discipline varies widely by team maturity. Separately, Dex Horthy's SlopCodeBench found unreviewed AI output contained 11 times more issues than reviewed code. Teams setting merge policies for agentic PRs can track this evolving benchmark data on daily.dev."}}]} ```