The rise of AI in healthcare is not just reshaping workflows—it’s upending long-standing professional hierarchies. Nowhere is this more evident than in the radiography-radiology divide, where the traditional delineation between ‘technicians’ (radiographers, sonographers) and ‘professionals’ (radiologists) is being turned on its head.
Ironically, the very work that was historically seen as the more ‘intellectual’ domain—diagnostics—is now the part most vulnerable to automation.
A Profession Built on Technology: The Origins of Radiology
Radiology as a profession emerged in response to the development of X-rays, a prime example of how technological innovation creates new occupational domains (for example, see Witz, 1992 Patriarchy and the Professions).
Yet, as the medical profession solidified its dominance, a division of labour was imposed:
- Radiologists took on the clean, diagnostic, interpretative work—reading scans, making clinical decisions, and holding professional autonomy.
- Radiographers and sonographers became the technical workforce—performing the hands-on imaging procedures but remaining in a subordinate role.
This split reinforced a hierarchy of expertise, with radiologists positioned as the gatekeepers of medical imaging interpretation while radiographers carried out the technical tasks.
The AI Disruption: When the ‘Cognitive’ Work Gets Automated
AI, particularly in radiology, is disrupting this hierarchy in unexpected ways. The work most at risk of automation is not the manual, technical component—it is the diagnostic pattern recognition that radiologists have historically owned. AI is proving to be:
✔ Faster and more accurate in detecting anomalies in medical images.
✔ Capable of reducing diagnostic backlogs, overcoming workforce shortages.
✔ Consistent and objective, avoiding human fatigue and bias.
This challenges the technicality-to-indeterminacy ratio—historically, professions have justified their autonomy based on their expertise in handling complex, unpredictable problems. Yet, if AI can outperform human radiologists in diagnostic precision, what remains of the profession’s claim to authority?
The New Skill Divide: What AI Can’t (Yet) Automate
Paradoxically, the roles least at risk from automation are the ones that have long been considered ‘technical’ rather than ‘professional’. Consider:
- Needle-guided biopsies: A highly complex, tactile task that requires real-time, three-dimensional navigation.
- Positioning patients for imaging: This requires expertise in anatomy, patient communication, and machine operation—elements that involve real-world, high-touch interaction rather than pure data processing.
- Ultrasound scanning: Sonographers don’t just ‘take images’—they interpret findings in real-time, adjusting angles, pressure, and positioning dynamically, an adaptability AI struggles with.
Thus, while AI can reduce radiological interpretation to an efficient, pattern-recognition task, the indeterminate, hands-on aspects of radiography remain deeply human-dependent.
Rewriting the Professional Hierarchy
Historically, professions have justified their existence and autonomy based on their control over complex decision-making. But as AI advances, we must rethink who is actually doing the “high-skill” work:
🤖 Radiologists’ work is being automated because it is reducible to rules and patterns.
🧑⚕️ Radiographers and sonographers remain indispensable because their work is highly context-dependent, tactile, and interactive.
This is an inversion of the traditional professional logic—one that challenges old assumptions about the division of labour in healthcare. The ‘intellectual’ work is being automated, while the ‘technical’ work is proving resilient.
Final Thoughts: The Professions at a Crossroads
The impact of AI on radiology is just the beginning. Across healthcare, we are seeing a shift where:
- Cognitive tasks are increasingly automated (diagnostics, pattern recognition, administrative decision-making).
- Hands-on, patient-facing work remains human-led (procedural expertise, communication, adaptability in real-world environments).
The question is: how will professional hierarchies evolve when the tasks once seen as ‘elite’ are the ones most easily automated?
The irony is clear: AI is automating the gatekeepers, while the hands-on experts remain indispensable.
Are we ready to rethink what defines professional expertise in an AI-driven world?
For a more detailed analysis of the gendered division of labor in the health workforce (, see:
Professions and Patriarchy (Witz, 1992)
The Sociology of The Professions (Macdonald, 1995)
Authors Tracey Adams and Ivy Bourgeault (multiple publications)
The Allied Health Professions: A Sociological Perspective. (Nancarrow and Borthwick, 2021)
