Today, Artificial Intelligence (AI) based technologies have the ability to not only reinforce long term structural bias (embedded in data) but also automate it in ways that are both invisible and unaccountable. In this project, we aim to highlight these inherent biases with the use of satirical design. We focus majorly on gender, age and language based bias in data and algorithms.
We started off by conducting an extensive literature review to understand bias in Artificial Intelligence and its sources. Next, we reviewed case studies of instances where bias in AI systems negatively impacted the marginalized population. We focused majorly on 3 case studies (1) Autonomous vehicles' bias against older adults, (2) Hiring systems' bias against women of color and (3) Voice recognition systems' bias against non-American accents. We designed posters branding hypothetical products that can help us navigate through these biases. The goal was to help the viewer realize the 'stupidity' of these 'smart' algorithms. Next, we focused on the second case study and made a video to satirically depict the background processing in a video interview assessed by an AI system.
Poster highlighting a product to fight bias in autonomous vehicles
Poster highlighting a product to fight bias in hiring systems
Poster highlighting a product to fight bias in voice recognition systems
1. Highlighting bias in Artificial Intelligence. Shubhangi Gupta, Raghav Mittal.