What Is Autocorrect? A Complete Definition

Autocorrect is software that detects and fixes spelling and grammar errors as you type, replacing incorrect words automatically. On Mac, built-in autocorrect uses dictionary lookup and basic language models. Modern tools like Charm add ML-based context awareness and grammar correction. Approximately 3 billion devices worldwide have some form of autocorrect enabled, handling an estimated 50 billion corrections every day.

How does autocorrect work technically?

Autocorrect operates in two stages: candidate generation and correction selection.

In the first stage, the system checks each typed word against a vocabulary. If the word is absent, it searches for close matches using an edit-distance algorithm - typically Levenshtein distance, which counts the minimum number of single-character insertions, deletions, or substitutions required to transform one word into another. This produces a ranked list of candidate corrections.

In the second stage, a language model re-ranks those candidates based on context. The simplest approach uses an n-gram model, which estimates the probability of a word given the preceding one or two words. More advanced systems use transformer models that process the entire sentence, producing significantly better results on ambiguous inputs where the correct word depends on context several words back.

macOS autocorrect is built on Apple's NSSpellChecker framework, present since Mac OS X 10.3. NSSpellChecker uses a vocabulary of roughly 10,000 common English words and a short n-gram context window. It fires on space or return, replacing the previous word if confidence is high. The critical architectural limitation: NSSpellChecker only works in apps that explicitly opt in to it. Native apps like Notes and Mail do; Electron-based apps like Slack, VS Code, and Discord do not.

What is the difference between autocorrect, spell check, and grammar check?

The three terms are frequently used interchangeably, but they describe distinct processes operating at different levels of language.

Feature	What it catches	When it runs	What it misses
Autocorrect	Misspelled words, transposed characters, fat-finger typos	As you type, on space or return	Grammar errors, correctly spelled wrong words (their/there)
Spell check	Words not in the dictionary	After you type, on demand or inline underline	Grammar errors, correctly spelled wrong words
Grammar check	Subject-verb agreement, tense errors, run-on sentences, punctuation	At sentence boundaries or on demand	Stylistic issues, contextual errors, domain conventions

The key distinction: spell check is passive and requires you to review flagged words; autocorrect is active and replaces without asking. Grammar check operates at the sentence level, analysing structure rather than individual words. Most Mac writing tools layer all three, but they use different underlying models.

How good is autocorrect on Mac?

macOS autocorrect performs well on short, everyday English words - the kinds of typos most people make most often. It struggles in predictable ways. Its 10,000-word vocabulary leaves large gaps for technical terms, proper nouns, and informal language. Its short context window means it cannot resolve ambiguity that depends on words more than one or two positions back in the sentence.

Research from Microsoft (2015) found that roughly 14% of autocorrect corrections on desktop keyboards are unwanted, with context failures accounting for the majority of false positives. That rate improves significantly as vocabulary and context window grow.

The coverage problem is equally significant. Because Slack, VS Code, Discord, and other Electron apps bypass NSSpellChecker, macOS autocorrect is simply absent for a large share of daily typing. Studies suggest the average Mac knowledge worker now spends over 40% of their typing time in Electron-based applications - meaning built-in autocorrect is inactive for nearly half of all keystrokes on a typical Mac.

What makes Charm's autocorrect different?

Charm's Spells feature operates via the macOS Accessibility API, giving it access to every text field on the system regardless of which framework the app uses. That includes Slack, VS Code, Discord, and any other app where built-in macOS autocorrect is absent.

The vocabulary difference is substantial: where NSSpellChecker covers roughly 10,000 words, Charm's ML model covers 100,000+ words, including technical vocabulary, common proper nouns, and modern informal language. The model uses a broader context window to evaluate corrections, reducing false positives on valid but unusual words.

When Spells makes a correction, the affected text briefly glows cyan - visible feedback that a change occurred, so you can review and revert instantly. Charm also adds two capabilities entirely absent from macOS autocorrect: Polish (grammar correction, blue glow, fires at sentence boundaries) and Oracle (word prediction, purple glow, Tab to accept). All processing runs on-device, with no text sent to external servers.

Frequently asked questions

What is autocorrect?

Autocorrect is software that detects and fixes spelling errors as you type, replacing incorrect words automatically. It uses dictionary lookup to identify misspellings and a language model to select the most probable correct replacement given the surrounding context.

Is autocorrect the same as spell check?

No. Spell check flags misspelled words and leaves the correction to you. Autocorrect goes a step further by replacing errors automatically without asking. Grammar check is a separate layer that analyses sentence structure rather than individual word spelling.

Does autocorrect learn from you?

macOS autocorrect adds words you repeatedly type to your personal word list but does not update its core language model. Charm stores accepted corrections locally to improve future suggestions, all on-device with no data sent to any external server.

Why does autocorrect make mistakes?

Mistakes come from three sources: vocabulary gaps (correct word not in the dictionary), context failures (limited context window cannot resolve ambiguity), and over-confidence (system corrects valid but unusual words). A larger vocabulary and deeper context window reduce all three error types.

How do I turn off autocorrect on Mac?

Open System Settings, go to Keyboard, select Text Input, and click Edit. Uncheck "Correct spelling automatically." This disables macOS autocorrect for native apps. Electron apps like Slack and VS Code manage their own text input and are unaffected by this setting.