--- title: AI Row Count Estimates Come From Nothing | daily.dev description: A playful cartoon-style post contrasts how database query optimizers estimate row counts (using real statistics objects, samples, and counts) versus how AI... canonical: https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: AI Row Count Estimates Come From Nothing | daily.dev og:description: A playful cartoon-style post contrasts how database query optimizers estimate row counts (using real statistics objects, samples, and counts) versus how AI... og:url: https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u og:image: https://api.daily.dev/og/posts/C8dTnsw4u.png og:image:alt: AI Row Count Estimates Come From Nothing 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. # AI Row Count Estimates Come From Nothing **[SQL Authority](https://daily.dev/sources/sql-authority)** · 2 min read · 33 upvotes · 5 comments ## Summary A playful cartoon-style post contrasts how database query optimizers estimate row counts (using real statistics objects, samples, and counts) versus how AI language models produce row count guesses (by pattern-matching plausible-sounding numbers from text, with nothing behind them). The core advice: when an AI gives a confident numeric estimate, ask it to point to something concrete backing the number; if it only offers reasoning without a source, it's a guess, not a measurement. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does a database query optimizer estimate the number of rows a query will return? A query optimizer builds its row count estimate from statistics it maintains on each column, including the number of rows and how values are distributed. These estimates can go wrong when the statistics are outdated relative to current table size, such as when a table has doubled since the stats were last collected, but this is a known problem with a known fix (updating statistics). _Developers tuning slow queries can compare optimizer statistics against AI-suggested estimates on daily.dev._ ### Why do AI models give confident but inaccurate row count estimates for SQL queries? AI models generate row count numbers based on how such answers typically sound in text they were trained on, not from actually examining any data, statistics, or samples. A number like four hundred seems plausible for almost any table, which is itself a sign the estimate has no real grounding. The way to test a model's estimate is to ask how it knows and see whether it points to an actual statistics object, sample, or count rather than just explaining its reasoning in prose. _Anyone verifying AI-assisted database answers can track these reasoning gaps through daily.dev before trusting a number._ ## Community discussion Top comments from developers on daily.dev. **@ajaychoudhary80** · 9 upvotes > Great analogy. A query optimizer has statistics behind its estimate; an LLM has patterns behind its guess. Those are very different kinds of evidence. **@ahmetozel** · 3 upvotes > The contrast with optimizer statistics is useful, although those statistics are still estimates with a provenance and an age. A number backed by a statistics object is inspectable; it is not automatically the current row count or the cardinality of a particular filtered query. > > I would ask an assistant to state which quantity is being requested before choosing the data source: total table rows, planner estimate, or exact predicate count. Attach the query and collection time to the result. That prevents a real but stale statistics value from receiving the same apparent certainty as a count... **@jenueldev** · 2 upvotes > The nice part is that optimizer estimates can be tested at plan review, even when they are wrong. A helpful AI integration could flag where estimated and actual rows diverge, then ask for the smallest missing signal instead of inventing a cardinality upfront. That turns it from a number generator into a troubleshooting guide. **@capestart** · 0 upvotes > Good reminder that confidence and access to the underlying data are two completely different things. **@ristotoldsep** · 0 upvotes > A confident guess is still a guess, it just has better posture. ## Similar posts on daily.dev - [AI Hallucinated a Table That Was Never There](https://daily.dev/posts/ai-hallucinated-a-table-that-was-never-there-z4oucb6lp) · SQL Authority · 87 upvotes · 10 comments - [AI Myths vs. Reality: What AI Can and Can’t Actually Do](https://daily.dev/posts/ai-myths-vs-reality-what-ai-can-and-can-t-actually-do-i5tm8jo2s) · Medium · 2 upvotes · 0 comments - [Making AI work for databases](https://daily.dev/posts/making-ai-work-for-databases-ismsqunbk) · InfoWorld · 0 upvotes · 0 comments --- Tags: [#database](https://daily.dev/tags/database), [#sql](https://daily.dev/tags/sql) [View this post on daily.dev](https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u) ```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":"AI Row Count Estimates Come From Nothing","url":"https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u"},"datePublished":"2026-10-03T02:04:53.435Z","dateModified":"2026-10-03T02:05:16.956Z","description":"A playful cartoon-style post contrasts how database query optimizers estimate row counts (using real statistics objects, samples, and counts) versus how AI...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0b07f9822855491bbe2a3a41d62cad0a?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0b07f9822855491bbe2a3a41d62cad0a?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"SQL Authority","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":"SQL Authority","logo":"https://media.daily.dev/image/upload/s--reCGnvgi--/f_auto,q_auto/v1780213436/logos/sql-authority?_a=BAMAMiWQ0","url":"https://daily.dev/sources/sql-authority"},"commentCount":5,"discussionUrl":"https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":33},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":5}],"keywords":"database,sql","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"SQL Authority","item":"https://daily.dev/sources/sql-authority"},{"@type":"ListItem","position":3,"name":"AI Row Count Estimates Come From Nothing"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u","comment":[{"@type":"Comment","text":"Great analogy. A query optimizer has statistics behind its estimate; an LLM has patterns behind its guess. Those are very different kinds of evidence.","datePublished":"2026-10-03T04:46:29.758Z","url":"https://daily.dev/posts/C8dTnsw4u#c-DFRDZLPgi","author":{"@type":"Person","name":"Ajay Choudhary","url":"https://daily.dev/ajaychoudhary80","image":"https://media.daily.dev/image/upload/s--WkX8wDW2--/f_auto/v1780989706/avatars/avatar_QNZlI2SgVnNSx5PSRpCVx?_a=BAMAMiWQ0"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":9}},{"@type":"Comment","text":"The contrast with optimizer statistics is useful, although those statistics are still estimates with a provenance and an age. A number backed by a statistics object is inspectable; it is not automatically the current row count or the cardinality of a particular filtered query.\nI would ask an assistant to state which quantity is being requested before choosing the data source: total table rows, planner estimate, or exact predicate count. Attach the query and collection time to the result. That prevents a real but stale statistics value from receiving the same apparent certainty as a count executed against the database state the user actually cares about.","datePublished":"2026-10-04T17:03:35.436Z","url":"https://daily.dev/posts/C8dTnsw4u#c-iz78vBKIV","author":{"@type":"Person","name":"Ahmet Özel","url":"https://daily.dev/ahmetozel","image":"https://avatars.githubusercontent.com/u/70992231?v=4"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3}},{"@type":"Comment","text":"The nice part is that optimizer estimates can be tested at plan review, even when they are wrong. A helpful AI integration could flag where estimated and actual rows diverge, then ask for the smallest missing signal instead of inventing a cardinality upfront. That turns it from a number generator into a troubleshooting guide.","datePublished":"2026-10-05T03:45:14.669Z","url":"https://daily.dev/posts/C8dTnsw4u#c-5eJdNkoEa","author":{"@type":"Person","name":"Jenuel Oras Ganawed","url":"https://daily.dev/jenueldev","image":"https://media.daily.dev/image/upload/s--BadmiNyj--/f_auto/v1787625077/avatars/avatar_JJtZr4Sjm?_a=BAMAMicg0"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2}},{"@type":"Comment","text":"Good reminder that confidence and access to the underlying data are two completely different things.","datePublished":"2026-10-07T10:41:37.823Z","url":"https://daily.dev/posts/C8dTnsw4u#c-sGoWc4tuW","author":{"@type":"Person","name":"CapeStart","url":"https://daily.dev/capestart","image":"https://media.daily.dev/image/upload/s--whj108mU--/f_auto/v1758187812/avatars/avatar_Pcn9ylzyt9IhD4SoHHPHt?_a=BAMAK+ZW0"}},{"@type":"Comment","text":"A confident guess is still a guess, it just has better posture.","datePublished":"2026-10-07T14:15:28.801Z","url":"https://daily.dev/posts/C8dTnsw4u#c-Qx05Ssm9z","author":{"@type":"Person","name":"Risto Tõldsep","url":"https://daily.dev/ristotoldsep","image":"https://lh3.googleusercontent.com/a/ACg8ocLDWc6mZn0JwNmXXw6WY0L_HJ6pegzRttooC5VgtXESHSj3MNYx=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/ai-row-count-estimates-come-from-nothing-c8dtnsw4u#faq","mainEntity":[{"@type":"Question","name":"How does a database query optimizer estimate the number of rows a query will return?","acceptedAnswer":{"@type":"Answer","text":"A query optimizer builds its row count estimate from statistics it maintains on each column, including the number of rows and how values are distributed. These estimates can go wrong when the statistics are outdated relative to current table size, such as when a table has doubled since the stats were last collected, but this is a known problem with a known fix (updating statistics). Developers tuning slow queries can compare optimizer statistics against AI-suggested estimates on daily.dev."}},{"@type":"Question","name":"Why do AI models give confident but inaccurate row count estimates for SQL queries?","acceptedAnswer":{"@type":"Answer","text":"AI models generate row count numbers based on how such answers typically sound in text they were trained on, not from actually examining any data, statistics, or samples. A number like four hundred seems plausible for almost any table, which is itself a sign the estimate has no real grounding. The way to test a model's estimate is to ask how it knows and see whether it points to an actual statistics object, sample, or count rather than just explaining its reasoning in prose. Anyone verifying AI-assisted database answers can track these reasoning gaps through daily.dev before trusting a number."}}]} ```