St. Luke's Health System

Justin Smith

The speed of innovation, along with accessible troves of data, is setting the stage for Machine Learning (ML) and Artificial Intelligence (AI) to catalyze business productivity. Despite seeing successful uses across industries, there persists a lack of understanding of ML and AI. Misunderstanding the capabilities and limitations of the technology poses a significant barrier to progress and adoption. Generally, individuals fall into two categories: –

A) ML/AI will solve ALL problems;

B) The technology is completely overhyped and worthless.

As with many situations, reality lies somewhere in the middle and education is the key to success.

Finding a middle ground through education is a powerful way to both increase understanding of the capabilities and limitations of ML/AI technology and establish trust between stakeholders and practitioners. One of the first steps in this process is to identify the target audience and prepare material to best suit their needs. At the executive level, focusing on downplaying ML/AI hype and highlighting examples within your organization can go a long way to winning champions.

Often, senior leadership may be familiar with a highly visible project or initiative that required ML/AI, but the connection that ML/ AI played in the process is lost. Education that contains initiative-level examples allows for tangible and memorable connections for strategic individuals within the enterprise to understand these new technologies and their efficacy.

At the operational management level, education on how to employ results from ML/AI solutions is critical. These individuals benefit from transparency in how a prediction was made, along with the explanation of what factors (features) drive the results. They want to know which levers they can pull to make actionable business decisions and drive change. Again, use cases that are specific to their department work best for this audience.

"Despite seeing successful uses across industries, there persists a lack of understanding of ML and AI. Misunderstanding the capabilities and limitations of the technology poses a significant barrier to progress and adoption"

Successful education campaigns must be tailored to the unique culture of each organization. Capitalizing on multi[1]modal venues for education delivery to match the audience is recommended. Two examples include: - 1) small in-person sessions delivering light technical exposure via use cases and the ability to foster a conversation between practitioners and stakeholders, 2) written articles published on the organization’s internal website or platform that describe ML/AI tools and technology, how they deliver solutions to specific problems and therefore drive actionable decision-making. Customizing education to match the organization’s culture cannot be overstated.

Over time, an army of ML/AI champions will emerge with an increased level of understanding.

Another recommendation to leverage these champions is to create a pathway of how to reach the data science team if they have questions. If possible, have an open[1]door policy or lunch-and-learn events to further explain ML/AI processes and results to the level of detail that matches audience interest. These types of discussions open the door to drawing attention to foundational areas of data and analytics that may not receive as much attention such as data governance, business intelligence, and master data management. Education, when done properly, can pay dividends up and down the analytics pipeline for the benefit of the organization.