Nothing for Us, Without Us; Breaking Unconscious Bias in Building Products
summary
A Speech-Text-Analytic app initially produces inaccurate captions for users with disabilities; community feedback prompts accessibility fixes and a feature that recognizes Nigerian native languages. Co-creation with medical practitioners on an AI diagnostic project exposes dataset bias risks and the need for domain expertise. The session names concrete mitigation steps—research methods, feedback loops, and inclusive design practices—shows a demo prototype, and lists practical pitfalls and tools used.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:08Unconscious bias in product dev
- 4:36Bias in AI health solutions
- 7:47Identifying and addressing bias
- 9:55Inclusive products and communities
- 11:39Mitigating bias in products
- 14:53Pitfalls in inclusive design
- 16:40Inclusive vs exclusive products
- 17:22Inclusive communities beyond tech
- 19:31Summary and steps
- 21:21Closing remarks
related talks
34mAI and Accessibility: the Good, the Bad, and the Bollocks
23mDesigning for Accessibility
33mEliminating Unconscious Bias Through UX Design
01h 26mInclusive & Ethical NLP Design Building AI for Everyone
49mUsing AI to Streamline Inclusive Design Without Losing Humanity
10m#BuildPropulsionLab at #DevoxxUK: @jbaruch with @antetmc2
48mAI and Disability: Progress or Exclusion?
57mCan Gen AI make the world Disability inclusive?