Perspectives on the use of Artificial Intelligence in Speech-Language Pathology

possibilities and limitations

Abstract

The incorporation of Generative Artificial Intelligence (GenAI) into healthcare has intensified, encompassing diagnosis, prognosis, telehealth, clinical decision support, and scientific production. In Speech-Language Pathology and Audiology, this process has occurred unevenly and still without specific national guidelines, despite advances in the automated analysis of speech, voice, language, and swallowing. This paper aims to critically synthesize the scientific literature and current institutional guidelines, discussing possibilities, limitations, and principles for the ethical and evidence-based use of AI in Speech-Language Pathology and Audiology. In assessment and diagnosis, machine learning models have shown promising performance in the classification of communication disorders, in addition to advances in videofluoroscopic swallowing studies. However, limitations persist, such as small sample sizes, low cultural diversity, and lack of external validation. In intervention, tools aimed at therapeutic personalization, monitoring, and telerehabilitation show potential, although they depend on professional supervision and integration with individualized plans. In education, AI represents a new paradigm, requiring attention to ethics, academic integrity, and technological literacy. International guidelines converge in stating that AI should act as a support tool, with human oversight, transparency, data protection, and rigorous validation. In scientific writing, its use is accepted as operational support, without replacing human authorship. AI constitutes a promising instrument to expand diagnostic, therapeutic, and scientific resources in Speech-Language Pathology and Audiology; however, it requires rigorous validation, algorithmic transparency, professional training, and ethical responsibility.