AI of the Beholder: How Surgical and Medical Specialties View Intelligent Technology

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision-Making

Background:

• Clinician conceptualizations of artificial intelligence (AI) reveal underlying assumptions about professional authority and decision-making.
• Understanding how different medical specialties frame AI research is crucial for identifying divergent professional norms.

Purpose of the Study:

• To examine if surgical and medical specialties frame AI differently in research abstracts.
• To determine if these framing differences reflect divergent professional norms regarding AI.

Main Methods:

• Analysis of 1561 AI-related research abstracts from high-impact journals (2019-2025).
• Classification of abstracts using a large language model (DeepSeek Reasoner) on human-AI relationship, professional autonomy, and decision control.
• Statistical analysis (Chi-square, logistic regression) to assess differences by specialty and publication year.

Main Results:

• Surgical abstracts more frequently framed AI as assistive (69.8% vs 54.9%) and explicitly addressed professional autonomy (73.5% vs 61.3%) compared to medical abstracts.
• Surgical abstracts also more often specified decision control (69.3% vs 58.6%).
• These differences remained consistent across the 7-year study period and were independently associated with surgical specialty in multivariable analysis.

Conclusions:

• Surgical and medical specialties exhibit distinct patterns in conceptualizing AI, reflecting established views on authority and expertise.
• These differing framings have significant implications for AI tool design, clinical implementation strategies, and healthcare governance.
• Recognizing these conceptual divides is vital for navigating the future of algorithmically mediated healthcare decision-making.