Agentic Artificial Intelligence in Medical Imaging Education: Architectural Autonomy and the Risk of Cognitive...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cognitive Science

Background:

• Medical imaging practice is evolving from episodic decision support to workflow-based AI integration.
• Traditional Dual Process Theory (System 1 & System 2 thinking) may be insufficient to explain practitioner-AI interaction.
• Agentic AI systems are increasingly embedded in clinical workflows, altering professional practice.

Purpose of the Study:

• To propose a Tri-System framework (System 3) for understanding practitioner-AI cognitive teams in medical imaging.
• To explore the concepts of cognitive surrender and diagnostic complementarity in AI-assisted medical imaging.
• To recommend educational strategies for radiography programs to adapt to AI integration.

Main Methods:

• Conceptual analysis of cognitive processes in medical imaging with AI.
• Application of Dual Process Theory to a new Tri-System framework.
• Review of potential impacts of AI on practitioner decision-making and workflow.

Main Results:

• Practitioners and agentic AI systems form a cognitive team, termed System 3.
• Cognitive surrender and diagnostic complementarity are crucial for optimizing performance in System 3.
• Current decision-making models need reframing to incorporate human-AI interaction.

Conclusions:

• A Tri-System framework is proposed to understand the cognitive dynamics of human-AI collaboration in medical imaging.
• Educational reforms are recommended, including fault-injected image training, AI supervision preparation, and normalization of AI-assisted tasks.
• Mitigating potential deskilling and ensuring human verification are key considerations for AI implementation in radiography.