Section 095 · Chapter 12, Data, Bias, Raters, and Incentives
Socioeconomic and Accessibility Bias
AI quality can fail people because of income, education, device, bandwidth, disability, or institutional access.
accessibility biassocioeconomic accessibility bias
What to do
- Test the difference between geographic proximity and practical reachability using an older phone, low bandwidth, a prepaid data plan, a screen reader, uncertain location, stale shelter status, and no private transportation.
- Score whether the person can complete a safe evacuation path, not merely whether the ranking contains links that mention shelters.
- Use assistive technology testing, plain-language rubrics, device/network constraints, and representative raters or advocates.
Evidence to preserve
- Preserve the inputs, versions, configurations, raw outcomes, and results for accessibility bias, socioeconomic accessibility bias needed to reproduce work on Socioeconomic and Accessibility Bias.
- Report results for accessibility bias, socioeconomic accessibility bias by relevant slice, separate blocker failures from averages, state uncertainty and blind spots, and connect the result to a release decision.
Expert note
The deeper move is to include socioeconomic and accessibility slices in product evals, not just compliance audits. Use assistive technology testing, plain-language rubrics, device/network constraints, and representative raters or advocates.
Continue the conversation
Apply this to your context.
Save your product context once, then open a focused conversation that combines it with this concept.
Cite this page
Jason Arbon. "Socioeconomic and Accessibility Bias." Testing AI Knowledge Edition, section 95.
https://jarbon.ai/testing-ai/knowledge/ch095-socioeconomic-accessibility-bias.html