Nicholas Ng, JD CHC
The first use-case
Non-technical managers can use tools like ChatGPT and Microsoft Copilot today to augment the systems already in place for employee development. In the world of ethics and compliance, ChatGPT or Copilot can generate fictitious scenarios based on real world regulatory compliance investigations or incidents. These scenarios can accelerate exposure to once rare case-types or core skills. An essential characteristic of these consumer-level AI tools is that they enable learning that is adaptable, or as in a recent case, spontaneous.
Imagine the following: a new employee has joined your team, and you want to establish a performance baseline on a couple of essential role-specific competencies. This way, you can more appropriately manage expectations, provide resources, and customize mentorship.
Compliance generalists often find themselves facilitating or chairing compliance committees consisting of a diverse group of stakeholders. The responsibilities of creating an agenda, writing meeting minutes, and issue spotting are at the heart of this work. As a team learning facilitator, I wanted to create a bundle of resources for myself and other compliance specialists seeking a refresher on compliance committee liaison duties. The following is what I wrote in a query to Microsoft Copilot:
“Generate a scenario for a fictitious hospital board meeting. I want descriptions of topics discussed, stakeholders present, and potential issues identified.”
The result was a well-structured meeting minutes document of a familiar feeling but entirely fictitious hospital board. A great start to be sure, but I needed to steer Copilot in the right direction for the purposes of the learning exercise. Defining the use case was critical here because I needed to not only identify the end goal, but also who the learner might be and where they might be in their learning journey.
In this case, the learners are professionals who require practice transforming a lengthy, on-the-fly narrative into a specific format. I needed the original narrative created by Copilot to be less structured and more verbose — and more human. I provided the following instructions:
“I would like the scenario to read like a narrative or set of notes. It needs to be 500 words.”
With this refinement, I was much closer to where I wanted to be. I also directed CoPilot with the following:
“Increase the verbosity of the narrative. Include more quantitative data.”
The resulting document met the objectives of the learning exercise (and ran against my expectations). I ultimately needed the AI tool to introduce imperfection and verbosity, not eliminate it.
Where I want to go next
Case studies enable compliance teams to improve their investigative processes through discussion on investigation plans, risk areas, and comparable cases. Generative AI, in its consumer-facing forms, might be able to provide immediate value in this context. I presented the following command to ChatGPT:
“I am responsible for leading a team of healthcare compliance professionals’ case study session. The session is 30 minutes and will review 3 scenarios relating to potential healthcare compliance issues. Generate these 3 scenarios. Each must be approximately 75 words. Do not analyze compliance issues, just provide the facts.”
ChatGPT generated the requested scenarios, the 2nd of which is displayed below:
In a mid-sized regional hospital, an internal audit discovered that a member of the procurement committee maintained financial ties with a medical supply vendor. Records revealed that several contracts were awarded without competitive bidding, bypassing established policies. The vendor provided equipment at discounted rates along with incentives. Reviews over the past six months indicated repeated interactions between the committee member and the vendor. Hospital leadership is now reviewing policies to address potential conflicts of interest.
The scenario is nearly usable for the team’s case study session. Like the meeting minutes exercise, this scenario needs a dose of imperfection:
“Remove investigative steps from these scenarios. More underlying facts.”
With one clarifying prompt, ChatGPT was able to get to where I was steering the conversation:
At Regional General Hospital, a member of the procurement committee maintains a financial relationship with a primary medical supply vendor. Several recent contracts for equipment and supplies have been awarded without competitive bidding. The vendor offers discounted prices and periodic bonuses, and correspondence shows regular interactions between the committee member and vendor representatives. Contract documents reveal pricing and delivery terms that differ from standard agreements. These arrangements coincide with multiple contracts finalized over the past few months and notable deviations in procurement practices.
This has been my start in using generative AI as an employee training and preparedness tool. I look forward to exploring how these tools can be used to augment ethics and compliance training workflows. Before implementing or advising on any AI-based workflows, it is strongly recommended that you consult your organization’s leadership and communications teams in addition to existing or developing policies.
Nicholas Ng, JD CHC, is a healthcare compliance specialist located in the Pacific Northwest. He is a true generalist, supporting primary care and specialty clinics as well as multiple tertiary care facilities. Nicholas enjoys learning about how ethics and compliance work can be a proactive partner with healthcare providers, industry stakeholders, and patients.