AI is changing faster than universities can govern it through traditional committees, policies, and curriculum cycles.
AI is starting to help with decisions about people's health, in hospitals, and in pro sports, where teams use data about players' bodies to decide who plays and when. Darwin by Medigram is the system that checks the AI: it scores every decision on six kinds of trustworthiness and keeps a permanent record, so whether the stakes are a patient's care or a player's career, nothing happens unchecked and there's always proof. Designed to work on mobile, where the work happens.
The provost’s challenge is not simply deciding where AI should be permitted. It is ensuring the institution can preserve academic and professional standards, define competent AI use, maintain accountable human authority, and demonstrate that consequential AI is operating within institutional boundaries.
A plain-language reader path into Building Darwin, the full CEO Letter.
The provost’s problem
AI now reaches teaching, assessment, research, administration, professional education, clinical environments, intellectual work, and institutional operations. A policy written for one context cannot automatically govern the others.
The central question is becoming institutional competence: what should students, faculty, researchers, professionals, and systems be able to do with AI, what remains constrained, who retains authority, and how does the institution know its expectations are actually being met?
Why universities keep getting this wrong
Universities distribute authority by design. Faculty governance, academic departments, research administration, IT, cybersecurity, legal, institutional research, professional schools, and central administration each hold legitimate pieces of the problem. AI crosses all of them.
Each group can perform its own work correctly while significant risks remain in the spaces between them. No committee or functional office automatically owns the complete path from academic principle through technical implementation, verification, evidence, and institutional accountability.
The problem is not necessarily weak specialists. The problem is unowned seams.
Owning the seams end to end
That gap between functions, not weak people within them, is where governance actually fails. This architecture exists to keep the full path (academic principle, implementation, verification, evidence) owned end to end rather than handed off department to department.
Why Sherri and the Medigram team
Darwin was architected across those same kinds of boundaries rather than passed sequentially among disconnected functions. Sherri Douville worked from accredited standards requirements through architecture, AI engineering, security, adversarial verification, evidence design, and implementation, so a finding in one layer could change decisions in the others. The lesson for higher education is not that one person should do everything. It is that someone has to own the seams, and the destination is infrastructure and operating discipline that make that integration repeatable, not dependent on one extraordinary individual holding it together.
Five questions for provosts
- What constitutes competent AI use in each discipline?
- Where must human academic or professional authority remain explicit?
- How does the institution distinguish AI assistance from unsupported or unverifiable work?
- Who owns risks that cross academics, research, technology, security, and administration?
- What evidence can the institution produce that its AI standards are actually operating?
From AI policy to AI readiness
Policies establish expectations. Readiness requires people who understand those expectations, systems that enforce appropriate boundaries, mechanisms that detect failures or drift, and evidence showing whether intended governance is working.
This does not require every university to become an AI engineering company. It requires academic leadership to know which responsibilities belong to the institution, which belong to technology providers, and where evidence is necessary to trust the boundary between them.
The goal
Higher education should be able to adopt AI without weakening the standards by which knowledge, professional competence, research, and institutional decisions earn trust. The opportunity is to make those standards more explicit and demonstrable as AI becomes part of everyday academic work.
Need the full architecture and evidence?
The CEO Letter contains the detailed technical rationale, verification loop, standards context and diligence path.