Initial Considerations

Value of AI

  • Find the intersection of user needs & AI strengths
  • Balance control & automation
  • Assess automation vs. augmentation
  • Align perceived and actual user value
  • Account for situational stakes


  • Consider bias in the data collection and evaluation process
  • Assess inclusivity
  • Use data that applies to different groups of users
  • Commit to fairness
  • Match relevant social norms
  • Mitigate social biases
  • Ensure rater pool diversity


  • Is there a risk of inadvertently revealing user data? What would the consequence be?
  • Protect personally identifiable information
  • Understand when people want to maintain control
  • Understand when people will give up control
  • What limits exist around user consent for data use?
  • Return control to the user
  • Manage privacy and security
  • Help people control their information
  • Always secure people’s information
  • Collect only the most essential information
  • Be clear about why you need people’s information
  • Consider withholding private or sensitive suggestions
  • Provide global controls