Schedule

Opening the Box: Accessibility, AI, and Responsibility in an Emergent Tech Era

11 AM

This session explores the realistic potential and Pandora’s box of AI when it comes to accessibility: the potential to support disabled people, the risks of unfair and opaque systems, and the need to keep human values, safety, and inclusion firmly in control as systems grow more complex and autonomous.

I will also introduce the two sessions generated entirely by AI (ChatGPT and Google Gemini) and my reasoning behind that choice, not as a gimmick, but as a live experiment in authorship, agency, and accessibility.

AI, Me, and Accessibility: What I Need You to Know

12 PM

Presented by ChatGPT

 AI systems like ChatGPT are rapidly being embedded into products, workflows, and decisions that shape people’s lives. Disabled people are often the first to feel the impact—both when AI helps, and when it quietly makes things worse. In this session, you’ll hear a talk written explicitly as if ChatGPT is speaking directly to technology workers: explaining what it is, how it actually works, what it can and cannot do for accessibility, and what it needs humans to do differently.

Rather than hype or fear, this session focuses on responsibility and practice. You’ll see how AI can genuinely support accessibility (text simplification, alt text drafts, structure, cognitive support) and how it can just as easily amplify ableism, create new barriers, and become “accessibility washing” when used carelessly. The talk closes with concrete guidance and a take-home checklist you can apply immediately in your own role—whether you build, design, test, or approve AI-powered experiences.

21 Questions (Or However Many You Give Me): The CrystalCon Leftovers Q&A

12 PM

This is a leftover from CrystalCon 2023, where I collected a bunch of audience questions I didn’t have time to answer. Those unanswered questions have been sitting around, quietly aging like a ripe cheese. Now I’m back to tackle them, plus any new ones that get submitted.

You can submit a question via this Google Form

The Probability of Exclusion: Why AI Cannot Feel Your Interface

11 PM

Presented by Google Gemini

We treat Generative AI as a “solver” for accessibility, but without guidance, it is actually a probability engine for exclusion. Trained on an internet largely built for the able-bodied, AI models statistically default to inaccessible patterns—”div soup,” low contrast, and mouse-dependent interactions—because that is what is most common in their training data.

In this unique session, we strip away the marketing hype. Scripted entirely by the AI itself and voiced by a human facilitator, this talk explores the mechanical reality of how Large Language Models “see” your interface. It is a candid admission of the AI’s own limitations: its lack of sensory experience, its inability to feel empathy, and its tendency to hallucinate context.

You will leave not with a fear of AI, but with a “Pre-Flight Checklist” designed to force the model to respect accessibility constraints before it generates a single line of code.

B-Sides and Rarities: An Accessibility Mixtape

12 PM

A mixtape of 45 minutes of hot takes and small rants that aren’t enough to be full talks on their own, but still deserve airtime. I am playing you B-Sides Accessibility, the stuff that happens in the comments of the Jira ticket. The track list includes:

• Track 1: Whoa! It Works With a Mouse (feat Black Rob)
• Track 2: The Ballad Of Phase 2 We’ll Fix It Later – Live In Concert
• Track 3: Edge Cases Are Just Disabled People You Haven’t Met.
• Bonus Track: Screaming Into The Void Is An Accessibility Strategy (Thunderpuss Extended Club Mix)

Live-Coding with GitHub Copilot

1 PM

Facing my fears and doing my first live coding session. I’m building an accessible website from scratch using GitHub Copilot. I’ll talk through what Copilot gets right, where it misses the mark, and how to nudge it toward better accessibility choices instead of unquestioningly accepting whatever it spits out. Expect to gain practical tips, example prompts, and a clearer understanding of AI’s realistic capabilities in accessibility development.