The reCAPTCHA verification period has expired. Please reload the page.

UX for AI World

Creating exceptional human-AI interactions.

Overview

As AI permeates our every interaction with technology, its form, shape, and contours in defining user experiences become more prominent. At HTC NXT, we believe user experiences have to place users as alpha and omega. We know that user experiences are too important to leave to chance, and our design teams leverage world-class practices to build, roll out, and use AI. That’s what we call Humane AIX. 

How we create humane AI experiences

How we create humane AI experiences

  • We start with identifying users and their experience expectations from ‘moments of truth.’

  • We acknowledge AI’s limitations and capabilities to serve these needs.
  • We apply the principles of the Humane AI Experience to extend the experience.

Humane AI design principles

blank

Humane AI design principles

  • Understand the design voice: AI is fundamentally moving technology from passive (user drives) to generative (AI decides based on user input) to intuitive (AI decides based on user behaviour) An aggressive AI design voice ignores user agency. A passive AI voice fails to exploit the AI opportunity.
  • Build a plan A: AI cannot do everything building a plan A involves
    • Understanding AI can’t solve problems (reliably) that cannot be conceptualized in design
    • Understanding where to automate where to augment and where to Abstain
    • First principles: start small and grow
  • Have a plan B: 
    • Degrade gracefully design with exit ramps from AI either to an alternative passive digital experience or an associate experience
    • Keep the user in content reverse bad steps (playback control) Feedback
  • Prioritize Trust:
    • Declare AI usage
    • Make AI Decisions explainable
    • AI has an overconfidence problem indicating confidence in AI results
    • Seek data permission. not capture forgiveness
    • Look at accuracy over
  • Build a continuous feedback loop:
    • Design in a Human-in-the-loop
    • Continuously hunt bias in performance data
    • Baseline UX research and scale

Our processes

blank