Embedded Files
Empower the user by creating trust and transparency in AI-powered digital products
Home / Leadership / Mayo Clinic / Case Study: AI UX Design Principles
Summary
The emergence of ChatGPT and AI-empowered applications seemed to explode almost overnight promising immediate off loading of tedious and repetitive tasks. The healthcare industry is an ideal place to focus this type of technical power but just like the days of the early internet, there was confusion with how to properly integrate and design for it. We needed to quickly figure out what design standards aligned best with AI technology so individual teams could continue to design for their portfolios quickly and autonomously.
Our Challenges
Product teams were inconsistent in how they defined AI and created AI-powered user interfaces.
Some teams abandoned best-practices in user interaction and usability all together to include generated content or functions.
Product leaders and stakeholders made rapid reprioritization to shifting work in progress to incorporate novelty AI chat messaging interactions.
Approach & Strategy
Established AI Principles Discovery Team
Created a cross-functional AI discovery team to review and inventory both internal and external instances of AI-powered applications.
Grouped and organized examples to highlight consistencies that aligned with best-practices in usability heuristics, ethics, and accessibility.
Developed AI UX Principles
Crafted a set of principles complete with extensive examples to be referenced by design and product teams.
Conducted a roadshow of the principles to help leadership and staff working on AI products to ensure.
Developed a roadmap for rolling out principles which included a larger AI adoption committee with the ultimate goal of sharing more broadly throughout the healthcare industry.
AI UX Design Principle Examples
Citations & References
It is critical when offering guidance to physicians to provide medical sources for further follow-up. In this example, a physician has asked about a medical condition and is provided a summary of parameters including related medical publications.
In this principle we state that you must provide sources and references when summarizing.
Highlighting Factors
There are many factors that might go into creating a health score or rating. Introducing a new score can be a challenge when physicians use many different factors and the correlation between them to determine a diagnosis or treatment plan.
In this principle, we highlight the importance of showing our work by explaining how a risk score or result is calculated.
Feedback
Physicians were very eager to explore and embrace AI-powered features in everyday tools, but there was a lack of trust in the accuracy of them. In this example the user does not like the response, so we've opened a dialog to understand more about this feedback.
In this principle example, we state that when ideas or content is generated, the user retains control by providing feedback to improve the model.
Google Sites
Report abuse