Vittorio

Design System Architect / Senior UX Designer

HAX Principles

Designing transparent, predictable AI experiences following Microsoft HAX guidelines

Convey the Consequences of User Actions

Help users understand how their actions affect the AI system and outcomes.

Governance

Encourage Granular Feedback

Enable users to provide specific feedback on AI performance, improving the system iteratively.

Governance

Learn from User Behavior

Adapt to user preferences over time, improving personalization while respecting privacy.

Adaptation

Make Clear How Well the System Can Do What It Can Do

Provide confidence indicators and accuracy information so users can make informed decisions.

Clarity

Make Clear What the System Can Do

Users should understand the AI's capabilities and limitations upfront, setting appropriate expectations.

Clarity

Make Clear Why the System Did What It Did

Provide explanations for AI decisions, helping users understand the reasoning behind actions.

Explainability

Match Relevant Social Norms

Design AI interactions that feel natural and appropriate within cultural and social contexts.

Alignment

Mitigate Social Biases

Ensure AI systems are fair, inclusive, and free from harmful biases.

Alignment

Notify Users About Changes

Keep users informed about system updates and changes that affect their experience.

Governance

Provide Global Controls

Give users control over AI behavior and settings at a system-wide level.

Governance

Remember Recent Interactions

Maintain context across user sessions, remembering relevant past interactions.

Adaptation

Scope Services When in Doubt

Limit AI actions when uncertainty exists, prioritizing user safety and control.

Recovery

Show Contextually Relevant Information

Present information that matters to the user's current task and context.

Context

Support Efficient Correction

Enable users to fix AI mistakes quickly and easily, with clear correction paths.

Recovery

Support Efficient Dismissal

Allow users to easily ignore or dismiss AI suggestions without friction.

Recovery

Support Efficient Invocation

Make it easy for users to access AI features when needed, with clear entry points.

Recovery

Time Services Based on Context

Deliver AI assistance at the right moment, when it's most useful and least disruptive.

Context

Update and Adapt Cautiously

Make changes that improve rather than disrupt user workflows, with careful testing.

Adaptation