--- title: ls AI Code Actually BAD? | daily.dev description: A hands-on exploration of whether AI-generated code is actually bad, using the Hazel game engine as a real-world test case. The author uses Claude Code to fix... canonical: https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: ls AI Code Actually BAD? | daily.dev og:description: A hands-on exploration of whether AI-generated code is actually bad, using the Hazel game engine as a real-world test case. The author uses Claude Code to fix... og:url: https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx og:image: https://api.daily.dev/og/posts/nd5yyskqx.png og:image:alt: ls AI Code Actually BAD? 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. # ls AI Code Actually BAD? **[The Cherno](https://daily.dev/sources/cherno)** · 55 min read · 5 upvotes · 4 comments ## Summary A hands-on exploration of whether AI-generated code is actually bad, using the Hazel game engine as a real-world test case. The author uses Claude Code to fix a deprecated Dear ImGui API call and then investigates a runtime physics crash caused by a kinematic body velocity calculation using wall-clock delta time instead of Jolt's fixed time step. The session demonstrates both the strengths of AI assistance (catching disabled code paths, systematic debugging, root cause analysis) and its limitations (needing human steering, inconsistent code style, requiring domain context). The core takeaway is that AI code quality correlates strongly with the engineering skill of the person directing it. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why would a kinematic body in a physics engine like Jolt start moving with runaway velocity at high frame rates above 60Hz? The bug occurs when code computes velocity using wall-clock frame delta time (real elapsed time between frames) instead of the physics engine's fixed time step, while the physics integrator still applies its own fixed time step to that velocity. At frame rates well above 60Hz, this mismatch causes kinematic body velocities to compound and grow exponentially each frame until they become non-finite, crashing the physics step. _Developers chasing similar physics desyncs can find real debugging walkthroughs like this one on daily.dev._ ### How reliable is Claude Code at fixing unfamiliar C++ codebase bugs without any prior project context? It successfully fixed a Dear ImGui deprecated API breakage (SetItemAllowOverlap removed in version 1.92) within minutes with zero prior context, even catching a duplicate instance of the same bug elsewhere that a human missed. However, on a harder NaN-triggered physics crash it initially proposed two incorrect root causes before needing added diagnostic logging and human steering to find the actual bug in the time-step handling. _Anyone evaluating AI coding assistants on real legacy codebases can track hands-on reports like this via daily.dev._ ### What percentage of AI-generated code is both functionally correct and secure according to Endor Labs? Only 10% of AI-generated code is both functionally correct and secure, according to Endor Labs, the maker of the AURI security tool. AURI is designed to catch vulnerabilities, block malicious packages, and write fixes directly inside Claude Code, Cursor, or VS Code, and is offered free for individual developers. _Teams shipping AI-generated code faster can follow security tooling coverage like this on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@harshitmanepally** · 0 upvotes > It depends on who you ask. **@ilijaveselinovic** · 0 upvotes > If agent can helps me, than it's not bad. **@haohmaru444** · 0 upvotes > if code is unuseable it means you can't prompt ## Similar posts on daily.dev - [On AI](https://daily.dev/posts/on-ai-y4nttoreo) · openmymind · 0 upvotes · 0 comments - [Reports of code's death are greatly exaggerated](https://daily.dev/posts/reports-of-code-s-death-are-greatly-exaggerated-3336mspgf) · Hacker News · 14 upvotes · 2 comments - [Bad Vibes Coding](https://daily.dev/posts/bad-vibes-coding-d9novql6h) · Lobsters · 1 upvotes · 1 comments --- Tags: [#ai-coding](https://daily.dev/tags/ai-coding), [#claude](https://daily.dev/tags/claude), [#c++](https://daily.dev/tags/c++) [View this post on daily.dev](https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx) ```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":"ls AI Code Actually BAD?","url":"https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx"},"datePublished":"2026-07-26T11:55:36.393Z","dateModified":"2026-09-13T21:31:56.806Z","description":"A hands-on exploration of whether AI-generated code is actually bad, using the Hazel game engine as a real-world test case. The author uses Claude Code to fix...","image":"https://i.ytimg.com/vi/s0oN6Tx5Xxo/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/s0oN6Tx5Xxo/sddefault.jpg","isAccessibleForFree":true,"articleSection":"The Cherno","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":"The Cherno","logo":"https://media.daily.dev/image/upload/s--167Cywjt--/f_auto/v1729843010/logos/cherno","url":"https://daily.dev/sources/cherno"},"commentCount":4,"discussionUrl":"https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":4}],"keywords":"ai-coding,claude,c++","timeRequired":"PT55M","video":{"@type":"VideoObject","name":"ls AI Code Actually BAD?","description":"A hands-on exploration of whether AI-generated code is actually bad, using the Hazel game engine as a real-world test case. The author uses Claude Code to fix...","thumbnailUrl":"https://i.ytimg.com/vi/s0oN6Tx5Xxo/sddefault.jpg","uploadDate":"2026-07-26T11:55:36.393Z","duration":"PT55M","url":"https://api.daily.dev/r/nd5yyskqx","embedUrl":"https://www.youtube.com/embed/s0oN6Tx5Xxo"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Cherno","item":"https://daily.dev/sources/cherno"},{"@type":"ListItem","position":3,"name":"ls AI Code Actually BAD?"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx","comment":[{"@type":"Comment","text":"It depends on who you ask.","datePublished":"2026-07-28T02:58:29.631Z","url":"https://daily.dev/posts/nd5yyskqx#c-0uRY34zuq","author":{"@type":"Person","name":"HARSHIT MANEPALLY","url":"https://daily.dev/harshitmanepally","image":"https://media.daily.dev/image/upload/s--O0TOmw4y--/f_auto/v1715772965/public/noProfile"}},{"@type":"Comment","text":"If agent can helps me, than it’s not bad.","datePublished":"2026-07-29T11:33:29.356Z","url":"https://daily.dev/posts/nd5yyskqx#c-OhCoQIAt6","author":{"@type":"Person","name":"Ilija","url":"https://daily.dev/ilijaveselinovic","image":"https://media.daily.dev/image/upload/s--GqTPoTzA--/f_auto/v1738848590/avatars/avatar_Nes9qpxMjBg8ZBR90Du7C"}},{"@type":"Comment","text":"if code is unuseable it means you can’t prompt","datePublished":"2026-07-31T07:14:23.370Z","url":"https://daily.dev/posts/nd5yyskqx#c-JmMLktyNZ","author":{"@type":"Person","name":"AGO","url":"https://daily.dev/haohmaru444","image":"https://media.daily.dev/image/upload/s--v-j2MPzl--/f_auto/v1741093503/avatars/avatar_tUmHBMHpGRR7sWE96RDHS"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/ls-ai-code-actually-bad--nd5yyskqx#faq","mainEntity":[{"@type":"Question","name":"Why would a kinematic body in a physics engine like Jolt start moving with runaway velocity at high frame rates above 60Hz?","acceptedAnswer":{"@type":"Answer","text":"The bug occurs when code computes velocity using wall-clock frame delta time (real elapsed time between frames) instead of the physics engine's fixed time step, while the physics integrator still applies its own fixed time step to that velocity. At frame rates well above 60Hz, this mismatch causes kinematic body velocities to compound and grow exponentially each frame until they become non-finite, crashing the physics step. Developers chasing similar physics desyncs can find real debugging walkthroughs like this one on daily.dev."}},{"@type":"Question","name":"How reliable is Claude Code at fixing unfamiliar C++ codebase bugs without any prior project context?","acceptedAnswer":{"@type":"Answer","text":"It successfully fixed a Dear ImGui deprecated API breakage (SetItemAllowOverlap removed in version 1.92) within minutes with zero prior context, even catching a duplicate instance of the same bug elsewhere that a human missed. However, on a harder NaN-triggered physics crash it initially proposed two incorrect root causes before needing added diagnostic logging and human steering to find the actual bug in the time-step handling. Anyone evaluating AI coding assistants on real legacy codebases can track hands-on reports like this via daily.dev."}},{"@type":"Question","name":"What percentage of AI-generated code is both functionally correct and secure according to Endor Labs?","acceptedAnswer":{"@type":"Answer","text":"Only 10% of AI-generated code is both functionally correct and secure, according to Endor Labs, the maker of the AURI security tool. AURI is designed to catch vulnerabilities, block malicious packages, and write fixes directly inside Claude Code, Cursor, or VS Code, and is offered free for individual developers. Teams shipping AI-generated code faster can follow security tooling coverage like this on daily.dev."}}]} ```