Related Experiment Videos
SCALES-AI: A Supervision- and Context-Aligned Entrustment Framework for Integrating Artificial Intelligence Into
Carl Preiksaitis1, Garrison Nord2, Joshua Davis3
1Department of Emergency Medicine Stanford School of Medicine Stanford California USA.
AEM Education and Training
|June 8, 2026
Summary
We developed SCALES-AI, a framework for AI in education, to guide decisions on AI tool autonomy. This ensures safe, equitable, and auditable AI integration in medical training.
Area of Science:
- Medical Education
- Artificial Intelligence in Healthcare
- Educational Technology
Background:
- Artificial intelligence (AI) is rapidly entering emergency medicine (EM) education.
- Existing frameworks for AI adoption in education are often too technical or abstract.
- This gap risks piecemeal AI adoption, potentially undermining training and amplifying bias.
Purpose of the Study:
- To develop a practical framework for determining appropriate levels of AI tool autonomy in medical education.
- To provide guidance for the safe, equitable, and auditable integration of AI in educational settings.
- To address the need for a structured approach to AI adoption in emergency medicine training.
Main Methods:
- Developed SCALES-AI (Supervision- and Context-Aligned Levels of Entrustment for AI in Education) through a three-phase process.
- Included literature synthesis, multidisciplinary workshops, and iterative refinement with national stakeholders.
- The framework rates AI tools on an entrustment scale linked to Foundational Principles and Dimensions of Trustworthiness.
Main Results:
- SCALES-AI provides a five-level entrustment scale (0-5) for AI tools in educational use, from 'not appropriate' to 'full autonomy for low-stakes tasks'.
- The scale aligns Foundational Principles (human-centered augmentation, evidence-based scalability, context-specific adaptation, ethical foundation) with Dimensions of Trustworthiness (Ability, Integrity, Benevolence, Equity).
- It specifies supervision intensity, audit cadence, and re-evaluation triggers based on AI tool changes or performance drift.
Conclusions:
- Programs should adopt graduated entrustment for AI tools, with explicit scope controls, supervision, and re-evaluation triggers.
- The SCALES-AI Checklist and Inter-Rater Calibration Template support implementation and reproducibility.
- SCALES-AI enables safe, equitable, and auditable AI adoption in health professions education while preserving human elements of training.