Description
Algorithmic systems designed to automate or support human decision-making are now deeply embedded in everyday life. However, because these systems operate within specific social and cultural contexts, they can reproduce and amplify existing forms of harm, particularly for marginalized communities, including LGBTQIA2S+ individuals. Prior work on algorithmic harms [4, 9] shows that such systems often inherit biases from their training data, leading to consequences such as misgendering [2], suppression of queer content [10], and discriminatory security practices [12].
Relying on platform-led reforms or regulatory interventions can have several limitations, including slow-moving regulatory processes [5], misalignment between industry practices and community needs [1], and the persistent risk of ethicswashing [8]. In response, community-driven coordination mechanisms are emerging as promising alternatives, leveraging the dependence of AI systems on user data and interactions. Approaches such as data leverage [11], algorithmic collective action [6], protective optimization technologies [7], and studies on folk theories of algorithms [3, 13] illustrate how collective strategies can redistribute governance power and center affected communities.
Despite this growing momentum, important challenges remain. Barriers related to awareness, accessibility, and empowerment continue to limit the adoption of these approaches, and it remains unclear how to design algorithmic collective action mechanisms that are both usable and genuinely beneficial for the communities they aim to support. This workshop will explore how research on algorithmic collective action can better connect with the needs of LGBTQIA2S+ communities.
The workshop is grounded in a prior work synthesizing evidence and strategies around algorithmic harms, resistance, and empowerment across academic literature, the general population, marginalized groups, and LGBTQIA2S+ communities specifically, together with early insights from an ongoing LGBTQIA2S+ community survey which assesses public awareness of algorithmic harms and their sense of empowerment regarding AI-based decisions. These materials will be presented as prompts to support collective reflection, rather than as definitive conclusions.
Format
Following a brief introductory presentation, participants will engage in structured small group discussions grounded in scenario cards derived from documented incidents. Additionally, to scaffold the discussions and make them more actionable, we will provide a set of complementary materials, including role cards (e.g., decision-maker, policymaker, decision subject), stakeholder priority prompts (to surface differing values and constraints), and response cards highlighting existing community-driven resistance techniques. Participants will also receive a short handout to help structure their analysis.
The activity unfolds in three phases:
Phase 01
Lived experience
Participants draw on their own perspectives and lived experiences to analyze a scenario, identify stakeholders, existing strategies, and barriers to implementation, while considering both optimistic and pessimistic outcomes.
Phase 02
Role play
Participants adopt assigned roles using the role cards and revisit the scenario, focusing on trade-offs, stakeholder priorities, and complementary strategies.
Phase 03
Synthesis
Participants collaboratively synthesize insights to identify and prioritize actionable directions for intervention.
Overall, the workshop aims to surface tensions, reveal blind spots, and encourage more holistic and actionable directions for algorithmic collective action.