Vibe Coding and Accessibility: Trust, Tools, and Verification in AI-Assisted Development
taught by: Natalie Tucker
Session Summary
Vibe coding, coined by Andrej Karpathy in 2025, is an AI-assisted development approach where developers guide tools like Cursor or Replit using natural language instead of writing every line of code by hand. As authorship shifts from typing syntax to steering AI, this session examines what happens to accessibility and how experienced practitioners must adapt to remain accountable for inclusive outcomes.
Description
Software development is undergoing a structural shift. With “vibe coding,” developers use natural language prompts to generate, debug, and refine code through AI tools. The friction of syntax decreases. Velocity increases. Barriers to entry lower.
But accessibility has never lived in syntax alone. It lives in intention, structure, testing, and verification.
When code is generated rather than written, responsibility doesn’t disappear, it becomes easier to blur. If we are not inspecting every line, how do we ensure semantic integrity? If AI generates ARIA, who verifies its correctness? If non-programmers can build functional apps, who safeguards usability for assistive technology users?
This session approaches vibe coding not as a novelty, but as a systems change in how digital products are created. Through live analysis, workflow modeling, and accessibility testing demonstrations, we will explore:
• How AI-assisted workflows reshape expectations of authorship and professional responsibility
• The specific accessibility failure modes emerging from AI-generated interfaces
• Practical verification strategies that preserve usability in high-velocity environments
• How to embed accessibility into prompts, feedback loops, and QA pipelines
• The cultural risk of confusing “it works” with “it is usable”
Participants will leave with a structured framework for integrating accessibility into AI-assisted development—one that prioritizes evidence over assumption and inclusion over convenience.
Practical Skills
- Participants will analyze how AI-assisted development changes accountability structures in accessibility practice and articulate their professional responsibility within these new workflows.
- Participants will craft effective natural-language prompts that embed accessibility requirements into AI-assisted development workflows and use structured follow-up prompts to critically evaluate AI-generated code.
- Participants will distinguish which aspects of accessibility verification can be reliably supported by AI tools and which require human judgment, assistive technology testing, and usability validation.