03 / Proactively Accounting for AI Error
A big part of AI x UX design is accounting for AI error to keep a positive user relationship. I sketched out a few error states / situation and how I would combat.
A | AI Confidence Levels
To keep AI honest and trustworthy, giving an accuracy score will build trust for the user (even if the prediction is wrong or inaccurate)
D | User Guidance
By asking “Is this content relevant for you,” gives the user an opportunity to seamlessly teach AI.
C | User Feedback Loop
By giving users an opportunity to give AI feedback, AI can improve and learn. The messaging is supportive and encouraging, promoting the user to continually give feedback.
B | Error Prevention
By giving an error prevention message, AI can help users have a more streamlined process.
E | Fallback Mechanism + Transparent Communication
Realistically, the AI won’t always work. Planning ahead by showing an alt route + friendly error message
04 / Rewarding for Accuracy
The AI will be rewarded (given a higher feedback score) from the following positive attributes:
Positive user feedback — higher patient confidence levels, self-improvement + user satisfaction
Active community engagement
Compliancy with privacy standards
Continuous improvement to medical standards and advancements
Quick responsiveness to user feedback