Keeping the Pathologist Competent in the Age of AI: Skill Shifting Risk and Ethical Error Accountability in AI-Assisted Diagnostic Pathology Restricted; Files Only
Duford, Mackenzie (Spring 2026)
Abstract
Artificial intelligence is increasingly being integrated into diagnostic pathology through tools that assist with image analysis, triage, classification, and workflow prioritization. Much of the ethical discussion surrounding these systems focuses on technical performance, such as accuracy, efficiency, and error reduction. This thesis argues that those measures are important but insufficient. The ethical problem posed by AI in pathology is not only whether these systems work, but how they reshape diagnostic work, professional competence, training, and responsibility.
The central claim of this thesis is that reliance on AI in pathology can create a risk of deskilling by shifting pathologists from active interpreters of evidence to supervisors of algorithmically structured workflows. This matters ethically because diagnostic competence is not reducible to correct outputs alone. It includes judgment under uncertainty, recognition of error, and the capacity to intervene responsibly when a diagnosis is contested, atypical, or wrong. If AI systems gradually erode those capacities while responsibility remains attached to individual clinicians, then accountability becomes increasingly misaligned with actual control.
At the same time, this thesis argues that AI should not be rejected simply because it introduces these risks. When an AI-enabled approach genuinely improves patient care under real clinical conditions, clinicians and institutions may have a prima facie obligation to adopt it as part of best available care. That obligation, however, is conditional. AI use is ethically justified only when it is integrated in ways that preserve meaningful human oversight, maintain professional competence, and distribute responsibility across the clinicians, institutions, vendors, and regulators who shape diagnostic outcomes.
Drawing on comparative analysis from radiology, aviation, self-driving systems, law, cybersecurity, and process control, this thesis develops an ethical framework for AI-assisted pathology and offers recommendations for training, workflow design, governance, and accountability. It concludes that ethical AI in pathology requires not only better tools, but institutions willing to preserve the human capacities that safe and trustworthy diagnosis still depends on.
Table of Contents
Table of Contents
Chapter 1: Introduction and Problem Framing 1
Chapter 2: Economic Incentives in AI Adoption 10
Chapter 3: What is Diagnostic Competence? 14
Chapter 4: Anatomical vs Laboratory Pathology 23
Chapter 5: Comparative Ethics on Automation 32
Chapter 6: What We Owe Patients in an AI-Mediated Clinic: A Normative Case for “Best 54
Available Care”
Chapter 7: Training, Competence Preservation, and Learning the Machine 61
Chapter 8: Governance, Licensing, and Responsibility 67
Chapter 9: Ethical Test and Recommendations 73
Conclusion 77
References 80
About this Master's Thesis
| School | |
|---|---|
| Department | |
| Degree | |
| Submission | |
| Language |
|
| Research Field | |
| Keyword | |
| Committee Chair / Thesis Advisor | |
| Committee Members |
Primary PDF
| Thumbnail | Title | Date Uploaded | Actions |
|---|---|---|---|
|
|
File download under embargo until 27 November 2026 | 2026-04-20 12:19:42 -0400 | File download under embargo until 27 November 2026 |
Supplemental Files
| Thumbnail | Title | Date Uploaded | Actions |
|---|