Why AI matters at this scale

The CUNY School of Medicine (CSOM) is a public institution within the City University of New York system, dedicated to training physicians with a focus on primary care and serving urban, underserved communities. It operates a combined BS/MD program and other graduate degrees, integrating rigorous academic coursework with clinical training across affiliated hospitals. As a large public entity (10,001+ employees system-wide), CSOM manages complex administrative, educational, and clinical data flows under significant pressure to demonstrate outcomes, equity, and stewardship of public funds.

At this scale and in the high-stakes sector of medical education, AI is not a luxury but a strategic lever. Large institutions generate vast amounts of data from student assessments, simulation labs, clinical rotations, and research activities. Manual processes struggle to extract actionable insights from this data, leading to inefficiencies and missed opportunities for intervention. AI offers the capability to personalize education at scale, optimize resource-intensive operations, and accelerate the research that underpins both curriculum and community health mission. For a public school, improving board pass rates, student retention, and research output through AI directly translates to greater impact per public dollar invested, enhanced reputation for attracting talent and funding, and better fulfillment of its social contract.

Concrete AI Opportunities with ROI

1. Adaptive Learning & Predictive Student Support (High ROI): Implementing an AI platform that analyzes performance across exams, simulations, and clinical evaluations can create personalized learning pathways. It can predict students at risk of falling behind or failing board exams months in advance, enabling targeted faculty tutoring and resource allocation. The ROI is clear: higher board pass rates improve school rankings and accreditation, boost student loan repayment rates, and directly support the mission of producing competent physicians. Early intervention reduces costly remediation and attrition.

2. Clinical Documentation Assistance (Medium ROI): Deploying AI-powered ambient scribes and NLP tools in teaching clinics reduces the clerical burden on medical students and faculty preceptors. This reclaims hours per week for direct patient care and teaching, enhancing the educational experience and clinic throughput. ROI manifests as improved student satisfaction, potential increase in clinical revenue due to more accurate and efficient coding, and reduced burnout among supervising physicians.

3. Research Data Augmentation (High ROI): Providing faculty and student researchers with AI tools for systematic literature reviews, patient cohort identification from electronic health records, and preliminary data analysis can cut project startup times by half. This accelerates grant proposal generation, publication rates, and interdisciplinary research. The ROI includes increased grant funding, enhanced institutional prestige, and more opportunities for student involvement in cutting-edge research.

Deployment Risks Specific to Large Public Institutions

Deploying AI at a large public university medical school carries unique risks. Data Integration Complexity is paramount, as data is often siloed between university student information systems, hospital EHRs (like Epic), and research databases, requiring costly and technically challenging interoperability projects. Regulatory and Privacy Scrutiny is intense, involving HIPAA, FERPA, and potentially IRB protocols, making rapid iteration difficult. Cultural Change Management across tenured faculty, administrative staff, and hospital partners can slow adoption, as top-down mandates are less effective in academic settings. Funding and Procurement Cycles are often slow and subject to public bidding processes, hindering the ability to pilot and scale agilely with innovative vendors. Finally, Algorithmic Bias and Equity risks are magnified given the school's mission; an AI tool that inadvertently disadvantages certain student demographics would be both ethically catastrophic and reputationally damaging, necessitating robust bias auditing frameworks.