Marketplace Tech AI's role in improving accessibility
Nov 28, 2025
In this insightful discussion, Taylor Arndt, a blind accessibility and AI advocate, shares her journey from facing challenges in accessing educational materials to becoming a coding expert. She emphasizes that AI has the potential to enhance accessibility but warns that without proper training data, it risks perpetuating exclusion. Taylor also contrasts 'partner coding' with 'vibe coding,' highlighting that many AI-generated platforms often overlook crucial accessibility features. Her advocacy stresses the importance of incorporating voices and experiences of people with disabilities in tech.
AI Snips
Chapters
Transcript
Episode notes
Self-Taught Fixes Led To Coding
- Taylor Arndt bought a portable scanner to read inaccessible paper homework she received in school.
- That practical fix led her to teach herself coding at 14 to solve accessibility problems for others.
AI Can Amplify Accessibility Problems
- AI can equalize access but also amplify inaccessibility when trained on poor examples.
- Taylor Arndt notes many AI-generated sites lack proper labels, headings, and form fields needed for accessibility.
Train Models With Curated Accessible Code
- Demand that frontier labs include people with disabilities and curate accessible coding examples for model training.
- This will teach AI to generate accessible patterns by learning correct, real-world implementations.