Stanford researchers are exploring whether AI could help fill the huge gap in speech therapy services for kids, since there aren’t nearly enough speech-language pathologists to meet the need. Right now, the big-name language models — even advanced ones like GPT-4 and Whisper — aren’t very good at spotting speech disorders in children, but the study found that when some models were fine-tuned, their accuracy noticeably improved.
That means AI could eventually take over some of the time-consuming parts of the job, like screening kids, transcribing speech, or tracking progress, which would free up therapists to spend more one-on-one time with students. The catch? The models still have accuracy issues, show bias (better with boys than girls, English over other languages, older kids over younger), and require tons of real-world speech data that’s hard to collect because of privacy laws. Even with those challenges, the researchers say the results are encouraging — not because AI is ready to replace SLPs, but because it might finally become a useful tool alongside them.
Read the full article here: https://news.stanford.edu/stories/2025/10/ai-speech-pathologists-language-services-children
Do you know soome dealing with a speech disorder? Then call Speak Easy Rehabilitation at 919.346.3350 to schedule an evaluation to day.
#speechtherapy #slp #speakeasyrehab
