Queer in AI

Last tended May 21, 2023 Edit Source

Queer In AI: A Case Study in Community-Led Participatory AI

• highlights representational and allocational harms to to brittle representations
• narrow conceptualizations of fairness assumes queer identities known, observable, measurable, discrete & static
• Participatory methods address limitations by involving users as co-designers
• Reflexivity in participatory methods encourages transparency during design
• participation should not be mediated by companies developing AI, shift decision-making & design power to marginalized groups

1. Participation and Decentralization
   • hierarchy poses problems:
     • fails to reflect diversity of community
     • high-profile organizers face targeted harassment campaigns
2. Participation and Intersectionality
3. Participation and Community Leadership
   • community informed: incorporate lived experience to guide research questions, data collection & data interpretation
   • community involved: decision making & research planning
   • community led: research seen not only as a process to create knowledge but also educated and mobilize action