--- title: The AI Competence/Judgement Gap | daily.dev description: Explores a widening gap between what AI lets people do and what they're actually equipped to judge. AI expands doing-proficiency across many tasks (code,... canonical: https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The AI Competence/Judgement Gap | daily.dev og:description: Explores a widening gap between what AI lets people do and what they're actually equipped to judge. AI expands doing-proficiency across many tasks (code,... og:url: https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56 og:image: https://api.daily.dev/og/posts/tg7BURD56.png og:image:alt: The AI Competence/Judgement Gap 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. # The AI Competence/Judgement Gap **[Itamar Gilad](https://daily.dev/sources/itamargilad)** · 10 min read · 15 upvotes · 3 comments ## Summary Explores a widening gap between what AI lets people do and what they're actually equipped to judge. AI expands doing-proficiency across many tasks (code, design, legal contracts) but doesn't raise judgment proficiency to match, creating a 'danger zone' where people can't tell if AI output is good. Cognitive biases like Dunning-Kruger make self-policing unreliable. Discusses using AI itself as a judge, citing Anthropic's finding that Claude Code's Auto Mode classifiers caught 89% of harmful changes versus 13% for human reviewers after observing 93% approval rates from developers experiencing 'approval fatigue.' Raises concerns about roles blurring (PMs coding, developers doing product work) and a near-future scenario where AI-mediated critiques from colleagues become disempowering. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is Claude Code's Auto Mode and why did Anthropic build it? Auto Mode is a set of classifiers Anthropic built into Claude Code to review and judge the agent's own suggested changes, and it is now the default mode. Anthropic introduced it after observing 'approval fatigue,' where developers using Claude Code approved 93% of suggested changes without close scrutiny. In testing, the classifiers caught 89% of harmful changes compared to just 13% caught by human testers. _Developers weighing how much to trust AI coding agents can track tooling shifts like this on daily.dev._ ### What is approval fatigue in AI coding assistants like Claude Code? Approval fatigue describes developers becoming cognitively worn down from reviewing AI-suggested code changes and defaulting to approving them without careful scrutiny. Anthropic observed developers using Claude Code approving roughly 93% of agent-suggested changes, a pattern that motivated building automated classifiers to catch harmful suggestions humans were missing. _Teams relying on AI agents for code changes can follow findings like this on daily.dev to keep review practices sharp._ ## Community discussion Top comments from developers on daily.dev. **@harshitmanepally** · 1 upvotes > Well I think it will be better to use AI to polish oneself than to rely on it entirely. **@agustinbarrientos** · 0 upvotes > I think every AI reviewer needs a named human owner when the team can't judge its reasoning. **@whlsk34** · 0 upvotes > Very interesting! This all assumes that you can't replace people with just warm bodies. The human will be making high-level decisions for some time (single-entrepeneur calling agents of agents is not here--no it's really not) so humans need to be making decisions, which means the bots need to explain technical decision to a non-expert so that they make the same decision the expert would. AI is not that good, and may never be, and the human probably never will be. ## Similar posts on daily.dev - [AI Makes Implementation Cheaper. Judgment Becomes More Expensive.](https://daily.dev/posts/ai-makes-implementation-cheaper-judgment-becomes-more-expensive--ojztzrujf) · Medium · 0 upvotes · 0 comments - [Something Big Is Happening](https://daily.dev/posts/something-big-is-happening-lgyfb7xtb) · Reid Burke · 44 upvotes · 6 comments - [AI is an extinction-level event for your rules of thumb](https://daily.dev/posts/ai-is-an-extinction-level-event-for-your-rules-of-thumb-fpwqv3x5j) · shipping bytes · 0 upvotes · 0 comments - [Above the API: What Developers Contribute When AI Can Code](https://daily.dev/posts/above-the-api-what-developers-contribute-when-ai-can-code-5fi2t6rld) · DEV · 54 upvotes · 2 comments - [Cybersecurity’s New Elite Skill: Human Judgment](https://daily.dev/posts/cybersecurity-s-new-elite-skill-human-judgment-doiq8jlkt) · Security Boulevard · 0 upvotes · 0 comments --- Tags: [#ai-coding](https://daily.dev/tags/ai-coding), [#claude-code](https://daily.dev/tags/claude-code), [#product-management](https://daily.dev/tags/product-management), [#code-review](https://daily.dev/tags/code-review) [View this post on daily.dev](https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56) ```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":"The AI Competence/Judgement Gap","url":"https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56"},"datePublished":"2026-08-12T12:32:01.866Z","dateModified":"2026-09-14T06:55:07.123Z","description":"Explores a widening gap between what AI lets people do and what they're actually equipped to judge. AI expands doing-proficiency across many tasks (code,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/dfe7922c34b389fa183e5460eefba243?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/dfe7922c34b389fa183e5460eefba243?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Itamar Gilad","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":"Itamar Gilad","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/282e7e2f414a480c824e03d0e11c07b8","url":"https://daily.dev/sources/itamargilad"},"commentCount":3,"discussionUrl":"https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":15},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":3}],"keywords":"ai-coding,claude-code,product-management,code-review","timeRequired":"PT10M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Itamar Gilad","item":"https://daily.dev/sources/itamargilad"},{"@type":"ListItem","position":3,"name":"The AI Competence/Judgement Gap"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56","comment":[{"@type":"Comment","text":"Well I think it will be better to use AI to polish oneself than to rely on it entirely.","datePublished":"2026-08-14T10:02:00.398Z","url":"https://daily.dev/posts/tg7BURD56#c-d1t19RrIQ","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"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}},{"@type":"Comment","text":"I think every AI reviewer needs a named human owner when the team can’t judge its reasoning.","datePublished":"2026-08-13T20:20:25.139Z","url":"https://daily.dev/posts/tg7BURD56#c-VpYAGaFFc","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"}},{"@type":"Comment","text":"Very interesting! This all assumes that you can’t replace people with just warm bodies. The human will be making high-level decisions for some time (single-entrepeneur calling agents of agents is not here–no it’s really not) so humans need to be making decisions, which means the bots need to explain technical decision to a non-expert so that they make the same decision the expert would. AI is not that good, and may never be, and the human probably never will be.","datePublished":"2026-08-14T17:38:22.944Z","url":"https://daily.dev/posts/tg7BURD56#c-8hRwlSpDD","author":{"@type":"Person","name":"whlsk34","url":"https://daily.dev/whlsk34","image":"https://media.daily.dev/image/upload/s--O0TOmw4y--/f_auto/v1715772965/public/noProfile"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-ai-competence-judgement-gap-tg7burd56#faq","mainEntity":[{"@type":"Question","name":"What is Claude Code's Auto Mode and why did Anthropic build it?","acceptedAnswer":{"@type":"Answer","text":"Auto Mode is a set of classifiers Anthropic built into Claude Code to review and judge the agent's own suggested changes, and it is now the default mode. Anthropic introduced it after observing 'approval fatigue,' where developers using Claude Code approved 93% of suggested changes without close scrutiny. In testing, the classifiers caught 89% of harmful changes compared to just 13% caught by human testers. Developers weighing how much to trust AI coding agents can track tooling shifts like this on daily.dev."}},{"@type":"Question","name":"What is approval fatigue in AI coding assistants like Claude Code?","acceptedAnswer":{"@type":"Answer","text":"Approval fatigue describes developers becoming cognitively worn down from reviewing AI-suggested code changes and defaulting to approving them without careful scrutiny. Anthropic observed developers using Claude Code approving roughly 93% of agent-suggested changes, a pattern that motivated building automated classifiers to catch harmful suggestions humans were missing. Teams relying on AI agents for code changes can follow findings like this on daily.dev to keep review practices sharp."}}]} ```