Google Docs significantly improving grammar correction, adding spelling autocorrect

In addition to Docs gaining Smart Compose, Google is significantly improving the grammar correction it introduced earlier this year in the G Suite app. It’s also gaining spelling autocorrect from Gmail.

The current system introduced in February uses machine translation, and the process can be thought of as “translating from the language of incorrect grammar to the language of correct grammar.”

At a basic level, machine translation performs substitution and reorders words from a source language to a target language, for example, substituting a “source” word in English (“hello”) for a “target” word in Spanish (“hola”).

A significant improvement leverages neural machine translation to “correct many more of the grammar mistakes you may make while writing.” For example, this can correct tense and identify homonyms. Errors are underlined with a squiggly blue line that you can right-click to accept changes.

Behind-the-scenes, Google is leveraging Tensor Processing Units to run two models for Grammatical Error Correction:

• The first method takes good sentences and makes them worse by automatically translating them to some other language and then back to English.
• The second method extracts source-target pairs from Wikipedia edit histories with a minimal amount of filtration.

These Google Docs grammar improvements will be available in the coming weeks. Meanwhile, Docs is soon adding spelling autocorrect with suggestions tailored to words that might be specific to a business. In Gmail, this capability works in real-time. 

For example, if your organization has an internal project name or acronym that’s commonly used, Docs will not only stop underlining such terms, but will also suggest corrections when these terms are misspelled.

According to Google, 100 million grammar suggestions are flagged each week, with the company positioning these improvements as helping boost productivity through AI. Additionally, Smart Compose has saved users from typing over 2 billion characters every week.

FTC: We use income earning auto affiliate links. More.