SLP/IO defines six core principles for responsible use of language models in speech-language pathology practice. If you already follow all of these instinctively, this section isn’t for you. It’s for the CF, the grad student, and the experienced clinician who just started using these tools last month and doesn’t have a framework yet.

01. AI supports thinking; it does not replace judgment

Language models can help organize, structure, and refine clinical writing. The clinician remains responsible for meaning, accuracy, and individualization.

02. Never include PHI in public tools

Compliance with HIPAA and FERPA is non-negotiable. No student names, DOBs, school names, or identifiable details in any public LLM.

03. Outputs must always be reviewed

No LLM output should enter an IEP, progress note, or legal document without professional review and revision.

04. Better documentation should still sound human

Polished language is not the same as good clinical writing. If a note sounds generic, it probably is.

05. Evidence-based practice ≠ polished wording

A well-formatted paragraph can still contain vague goals, weak measurement criteria, or generic descriptions that fail the student.

06. Precision in language is not pedantry

Subtle shifts in wording can change clinical intent. Specificity protects students and their access to services.