top of page

Healthcare Leadership After the AI Pilot: Who Owns the Decision When Technology Changes the Workflow?

Writer: Justine Asante
Justine Asante
Nov 4
5 min read

AI in healthcare is moving beyond the pilot stage. Ambient documentation, predictive analytics, clinical decision support, scheduling tools, and generative AI are increasingly entering everyday operations.


That shift creates a leadership question many organizations have not fully answered: who owns the decision when technology changes the workflow?

The question is not simply who approved the software. It is who is responsible for the new sequence of work, who can intervene when the tool performs poorly, and who answers for the outcome when responsibility is distributed across executives, clinicians, informatics teams, vendors, and frontline managers.

Why healthcare AI governance must move beyond the pilot

A pilot can be bounded. A production workflow is not. During a pilot, leaders may track accuracy, usability, adoption, and user feedback. Once the tool becomes part of routine care, however, it changes who sees information first, which tasks are performed automatically, what gets escalated, and how much discretion remains with the person doing the work.


The World Health Organization’s 2024 guidance on large multimodal models identifies risks including inaccurate outputs, automation bias, privacy concerns, inequity, and unclear accountability.


The National Academy of Medicine’s 2024 AI Code of Conduct discussion draft similarly emphasizes that health AI must be accountable, transparent, safe, equitable, secure, and adaptive. These principles point to an important operational conclusion: AI governance is not only a technology function. It is a leadership and organizational-design function.


The American Medical Association’s principles for augmented intelligence also stress oversight, transparency, privacy, bias mitigation, disclosure, and liability. Together, these sources establish the outside research basis for responsible deployment. The interpretation offered here is more specific: the central leadership risk appears when an organization adopts the technology but fails to redesign decision ownership around it.


The workflow changes before the job description does


When an AI tool is introduced, the formal reporting structure may remain unchanged while the practical work changes immediately. A nurse may now review an alert generated by a predictive model. A physician may verify an ambient note. A manager may be expected to monitor adoption and address staff concerns. An informatics team may monitor system performance. An executive may be accountable for the business case.


Each person has a role, but a role is not the same as decision ownership. If the tool produces an incomplete summary, who decides whether the workflow continues? If staff override recommendations frequently, who determines whether the problem is training, configuration, or poor fit? If a bias appears only in one patient population, who has the authority to pause deployment?


Without explicit answers, accountability can dissolve into a chain of assumptions. Leaders assume the vendor validated the model. Clinicians assume the organization approved the workflow. Managers assume the technology team owns the risk. The vendor points to appropriate-use guidance. No single person or group owns the result as a whole.


What this means for healthcare leadership


For CEOs, COOs, CNOs, CMIOs, and CHROs, the issue is larger than compliance. Poorly designed AI workflows can create rework, increase cognitive burden, weaken trust, and place frontline leaders in the position of enforcing decisions they did not help design.


This is especially consequential for nurse managers and department directors. They are often closest to the operational effects of a new tool but furthest from the authority to change it. They may be measured on throughput, documentation completion, staffing efficiency, or quality indicators while having limited control over the technology, staffing model, or implementation timeline.


The Institute for Sustainable Authority™ interprets this as a form of responsibility centralization: the organization depends on a small number of leaders to absorb ambiguity, resolve exceptions, and restore performance after the formal implementation team has moved on. In that model, technology may be distributed, but authority remains concentrated.


The result is executive dependency rather than sustainable leadership capacity.


Four actions executives can take after the pilot


1. Assign ownership at the workflow level

Name the executive sponsor, operational owner, clinical owner, technical owner, and frontline escalation path. Then document who can approve changes, pause the tool, modify the workflow, and communicate a decision. A committee can advise, but it should not become a substitute for accountable ownership.


2. Redesign accountability around the new work

Review job expectations, performance measures, escalation protocols, and staffing assumptions after implementation. If a manager is responsible for adoption, provide time, authority, and access to the data needed to influence adoption. Do not add AI oversight to a role without removing or redistributing other work.


3. Build intervention points, not just training

Training helps people use a tool. Governance must help them challenge it. Create clear mechanisms for reporting inaccurate outputs, unsafe recommendations, workflow friction, inequitable effects, and unintended workload. Define how quickly concerns are reviewed and who makes the final decision.


4. Measure organizational effects, not only technical performance

Accuracy and utilization matter, but they are incomplete measures. Track rework, overrides, delays, staff confidence, patient communication, equity indicators, and changes in decision time. A tool that performs well technically but increases hidden work may not be creating sustainable value.


5. Treat frontline participation as decision infrastructure

Frontline clinicians and managers should participate before, during, and after implementation. Their role should not be limited to testing usability. They need a recognized voice in defining acceptable risk, workflow exceptions, success measures, and stop conditions.


The question leaders should ask now


Healthcare AI governance becomes meaningful when leaders can answer three questions without hesitation: Who owns the decision? Who has the authority to intervene? Who is responsible for learning when the workflow does not produce the intended result?


These are not merely technical questions. They reveal whether an organization has built distributed responsibility or simply added another layer of work to already constrained leaders. Technology can accelerate a sound operating model, but it can also expose an organization that relies on informal heroics, unclear authority, and executive rescue.


Leadership challenges rarely exist in isolation. When decisions, accountability, and operational responsibility continue returning to the same leaders, the issue may require more than another training initiative.


The Institute for Sustainable Authority™ works with organizations and senior women leaders to strengthen decision ownership, leadership effectiveness, and sustainable performance. Explore Organizational Advisory Services or Executive Leadership Coaching to examine the changes your leadership team needs.

References & Resources

1. Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models — World Health Organization — January 18, 2024. https://www.who.int/publications/i/item/9789240084759

2. Artificial Intelligence in Health, Health Care, and Biomedical Science: An AI Code of Conduct Principles and Commitments Discussion Draft — National Academy of Medicine — April 8, 2024. https://doi.org/10.31478/202403a


3. AMA Principles for AI Development, Deployment, and Use — American Medical Association — 2023, updated policy context in 2024. https://www.ama-assn.org/system/files/ama-ai-principles.pdf


4. Artificial Intelligence Risk Management Framework (AI RMF 1.0) — National Institute of Standards and Technology — January 26, 2023. https://doi.org/10.6028/NIST.AI.100-1

 
 
The Institute for Sustainable Authority | Healthcare Organizations & Senior Women Leaders Coaching

Executive leadership advisory, coaching, and organizational support for healthcare leaders moving through transition and expanded responsibility.

QUICK LINK

Contact

(628) 272 3495

  • Facebook
  • Linkedin

The Institute for Sustainable Authority™ provides nonclinical executive coaching, leadership development and organizational advisory. These services are not psychotherapy, nursing practice consultation, medical care or mental health treatment and do not create a therapist-client relationship. Results vary.
 

Copyright © 2026 Justine Asante. The Institute for Sustainable Authority LLC. All rights reserved.

bottom of page