HOA 225: Wrangling Auto-Correct

In this episode of Hands On Apple, host Micah Sargent tackles the perennial frustration of iPhone and iPad autocorrect, offering practical solutions to make it work better for users. He explains how autocorrect leverages a local dictionary, personal typing history, and a device-based transformer language model introduced in iOS 17, which improves context awareness. The episode emphasizes that rejecting incorrect suggestions by tapping the 'X' in the prediction bubble is a powerful way to train the system, and that inline predictions (activated by the spacebar) can significantly reduce typing effort. Sargent also covers the limitations of autocorrect, including its struggles with multilingual typing, names, jargon, and the inability to manually add or remove words from the dictionary. He highlights text replacement as a powerful but underused tool for creating custom shortcuts that sync across devices, enabling users to fix persistent autocorrect errors and streamline repetitive inputs like email addresses and boilerplate text. While acknowledging Apple’s lack of granular control over the dictionary, Sargent concludes that with intentional use and a few setup changes, autocorrect can become a helpful assistant rather than a hindrance.