The controversy around AI's role in accessibility is explored, questioning whether it can truly replace human expertise.
An overview of permanent disabilities like physical or neurodiverse conditions and temporary ones like injury or situational impairments is provided.
As populations age, developers must account for deteriorating vision, hearing, and cognitive functions by supporting zoom and using simple language.
A simple "Hello World" example demonstrates how AI tools like ChatGPT produce inaccessible code unless explicitly prompted for accessibility.
An initial review of a sample web application reveals multiple accessibility issues, such as low contrast and lack of keyboard navigation.
GitHub Copilot's chat functionality is used to target a specific problem like keyboard navigation, demonstrating how AI can teach developers solutions.
Prompts can be framed around user personas, like an elderly user, to generate broader accessibility improvements beyond specific code fixes.
After asking AI to "make the page more accessible," the results show improvements but also highlight remaining flaws that require human oversight.
The effectiveness of using AI for accessibility hinges on providing clear, specific, and well-framed questions to get the desired outcome.
Sensory-Minds GmbH
Offenbach am Main, Germany
56:00Karl Groves
56:01Dirk Ginader
30:07Julia Undeutsch
55:32Hidde de Vries
31:39Rainer Stropek
51:09Léonie Watson, Artur Ortega, Karl Groves & Geri Reid & Sara Joy
29:59Konstantin Tieber
1:06:13Chris Heilmann & Daniel Cranney & Dirk Ginader
Jobs that call for the skills explored in this talk.
Ironhack
Municipality of Madrid, Spain
Jordan Martorell S.L.
Barcelona, Spain
