Designing AI that Earns Trust: A UX Approach

AI is everywhere nowadays, from healthcare and finance to transportation and education. This integration has brought many benefits to our daily lives, but it also raises concerns about the trustworthiness of these systems. Trustworthy AI is essential to ensure that these systems make ethical and responsible decisions. They must be consistent with human values and not cause harm to individuals or society as a whole.

So, how can we design AI that is credible and reliable for users?

Building initial trust in AI

In order to win user acceptance, trust is an absolute must when it comes to AI. The representation of an AI, the ability to transparently explain its behavior and decisions, and the opportunity to try it out are important aspects for initial trust building. Here are some key principles:

• Representation and perception: To give users a positive first impression, use a familiar design with positive values. The design should be kept simple and well-structured.
• Step-by-step onboarding: Introduce users to the system step-by-step and introduce additional features at a time when they are needed. This will help users feel more comfortable with the AI and its capabilities.
• Consideration of mental models: If the AI procedure or parts of the AI are too complex to understand, communicate a simplified or metaphorical mental model instead. This will guide users to understand the AI and its capabilities better.
• Testability: Allow users to try out features of the AI in a low-risk environment. This will help build their confidence in the AI and increase their trust.

Developing continuous trust in AI

Once users have developed initial trust in the AI, it is equally important to maintain that trust through consistent measures. Here are some key principles for creating continuous trust in AI:

• Understand how the system works: Promote a general understanding of how the AI is built and how and with what confidence it makes decisions. This will promote greater transparency and accountability.
• Traceability through feedback of the system state: Users should be consistently informed of the system’s current state and what actions are impacting it in the background. This also promotes transparency and helps users comprehend the AI’s decision-making process.
• Promotion of the purpose and added value of user data and feedback: Motivation for providing this information should be fair and attractive to users. Users should also understand how their data is being used and why it’s important.
• Effective and responsible use: Don’t take away tasks that users enjoy. Tasks that are unsafe and unpleasant for users, on the other hand, can be handed over to an AI. This will help users see the value in the AI and trust it more.
• User control: Give users the ability to customize the AI output to their needs, edit it, or disable it. By doing this, users will feel more empowered and in charge.
• Consideration and handling of errors: Whether it is an error or an output perceived as erroneous due to the wrong context; provide users with a path to recover from the error.
• Fairness and Diversity: Ensure equal opportunity for all user groups through responsible data collection and preservation, data diversity, and the good intent of the AI system for individuals and the community. This will help users trust that the AI is working in their best interest and is not biased in any way.

Building and maintaining trust in AI is not just a matter of user satisfaction, but also of businesses’ success. By prioritizing trustworthy AI products, companies can establish themselves as leaders in innovation and earn a reputation for putting their users first. So let’s strive for AI systems that are transparent, reliable, and open, and that promote fairness and diversity, to build a more promising and trustworthy future.

This article was written for Sclable’s blog on Medium by our UX Architect, Doris Gnong-Müller. Follow us on LinkedIn to get notified of new posts or check out our website to see the work we do!

If you liked it, give it a 👏 and share if you ❤️.