Smartphone Predictive Text Quality Declining, Users Report; AI Integration Cited as Likely Cause

Smartphone users across platforms including Apple and Samsung have documented numerous complaints about deteriorating predictive text performance in recent years, ranging from autocorrecting correct words to incorrect ones, failing to learn from corrections, and inserting nonsensical character sequences. A PhD candidate studying generative AI's impact on writing notes a "demonstrable decline" coinciding with 2023 updates when Apple introduced on-device machine-learning transformer models to replace its deterministic system used since 2007, with Samsung following a similar pattern. While a University of Queensland expert acknowledges "considerable anecdotal evidence" of perceived decline, he cautions that objective deterioration is difficult to establish given systems have become more sophisticated overall. The shift from dictionary-based and statistical approaches to neural language models designed to infer context and user intent represents a fundamental trade-off: modern systems gain flexibility and contextual understanding but produce more variable outputs that can appear less intuitive to users, with errors becoming more noticeable.