AI-Powered AAC Apps

Augmentative and alternative communication supports people who communicate using methods in addition to or instead of speech. AAC may include gestures, sign, communication boards, symbol systems, text, and speech-generating devices. A person may use different methods across settings, and AAC is not limited to one diagnosis or to a tablet app. Word prediction offers candidate words based on entered letters or context. Phrase prediction and generative AI may offer longer suggestions, but more automation creates added risks: a suggestion can change meaning, tone, or intent. The user should be able to inspect, edit, reject, or disable suggestions and retain their own vocabulary and voice. Predictions should not be spoken or sent without the person’s clear selection. Research on word prediction has studied communication rate, including a 2007 experiment with pseudo-impaired participants rather than AAC users with disabilities. Its results therefore do not prove the same speed or effort benefit for all AAC users. A meta-analysis of 114 single-case AAC intervention studies with school-aged individuals with autism and/or intellectual disability found positive average outcomes but substantial heterogeneity; it did not test generative-AI phrase prediction across all AAC users. An AAC system should be selected with the user and communication partners, with support from qualified speech-language professionals when available. Consider access method, motor and vision needs, language, vocabulary, cost, offline operation, voice options, and privacy. Keep low-tech and alternative access options available. AI can offer candidate language, but the user remains the communicator and author of their message.