Recent developments in generative AI open the possibility of new approaches that could profoundly impact quality and rate of speech for people using AAC. ChatGPT, Bard, and other Large Language Models (LLMs) can generate text that is natural and relevant, with a high degree of control over communication style, phrasing, and verbosity – even to the extent that it can deliberately generate culturally relevant vernacular or slang.

Clearly, the application of AI technology to AAC software is relevant and offers potential to deliver a substantial improvement to quality and rate of communication. However, because the use of generative AI, and specifically LLMs, in AAC is in its infancy, it is important to consider its potential benefits as well as pitfalls.

This presentation will highlight areas in AAC that may be positively or negatively influenced by AI including Agency, Communication Rate, Contextual Relevance, Cognitive and Language Development, Personality, and Privacy.