AI and Accessibility

In the rapidly evolving landscape of technology, Artificial Intelligence (AI) stands out as a game-changer our industry. But what are the implications for accessibility and building compliant and inclusive experiences?

In this talk, we will discuss AI and accessibility in all its nuances. From significant advancements that improve experiences for all users to major concerns around the ethics and organizational investments in accessibility, we will cover it all. Either way, AI is here and we got a lot to talk about to ensure accessibility remains at the forefront of our work.

Transcript

1. AI & ACCESSIBILITY WE GOT A LOT TO TALK ABOUT
2. I AM NOT AN AI EXPERT I AM AN ACCESSIBILITY EXPERT
4. WHY? Because everyone asks…
9. WHAT IS AI?
10. AI “A simulation of human intelligence by a machine”
12. General AI Narrow AI Superintelligent AI
13. NARROW AI Type of AI that performs specific tasks and doesn’t learn beyond that.
17. GENERATIVE AI An extension of Narrow AI, with the purpose of generating content based on large amounts of data that it has trained on.
19. NOW… LET’S TALK ABOUT ACCESSIBILITY
20. ACCESSIBILITY Simply means the “ability to access”
23. AI ACHIEVEMENTS IN ACCESSIBILITY
24. MAKING WRITING EASIER ChatGPT Creating meeting summaries. Text to speech applications.
25. VOICE RECOGNITION Improvements to captioning and translation tools Improvements to voice command applications. Voice Access Voice Control
26. IMAGE RECOGNITION Image recognition and captioning applications. Advancements in apps for visually impaired users Be My Eyes SeeingAI
28. AND MORE! Improvements to screen reader interaction. AI and accessible code suggestions. And more!
29. CONCERNS WITH AI
30. AI DOES NOT REPLACE KNOWLEDGE EXPERTS
31. DESIGNERS & CONTENT WRITERS Writing accurate alt text. AI can guess, but it rarely replaces professionals.
32. ENGINEERS Still need to know how to properly construct websites. Still need to write compliant and properly structured HTML. AI can’t understand nuances of new custom widgets
34. HARD SITE
35. CHALLENGES WITHIN AI
36. BIAS IN THE SYSTEM Amplify existing biases in society Race Gender & Disabilities Users treated differently, and not for the better.
37. PRIVACY Re-identification of users in protected groups. Proliferation of synthetic data. We need a stronger framework to protect participants.
38. ERRORS AI’s confidence level: Can’t convey its limits. Accuracy of results. Users overly trusting AI results
39. AI ISN’T THE ANSWER… SO, WHAT IS?
40. WHAT CAN YOU DO? ENGINEERS
41. CLEAN UP YOUR HTML Fix incorrect HTML tags & attributies Focus on writing clean HTML
42. ACCESSIBILITY BROWSER TOOLS This Photo CC BY Axe-Core Browser Extension. Accessibility Insights for Web. Lighthouse
43. AUTOMATED TESTING Add automated accessibility scans into your CI/CD pipeline: axe-core lighthouse-cli Selenium from Deque.AxeCore & More
44. WHAT CAN YOU DO? ORGANIZATIONS
45. COMPLIANCE Identify and understand your compliance level
46. HIRE ACCESSIBILITY EXPERTS IN Engineering Project Management Design User Research Testing
47. LISTEN TO YOUR USERS Gather Feedback Run user studies
48. OVERALL: USE AI STRATEGICALLY
49. AI IS A TOOL So, use it! Be aware of it’s limits. Thoroughly vet the results.
50. AI ISN’T THE MAGIC ANSWER
51. RESOURCES AI and Accessibility: A Discussion of Ethical Considerations by Meredith Ringel Morris Impacts of AI on Web Accessibility by Ana Sofia Gala The Future of AI and Web Accessibility by ADA Site Compliance bemyeyes.com speechify.com
52. THANK YOU! Cat Johnson catandthemachines@gmail.com www.catandthemachines.com