Session: 2 for 1: Case Study: Building Accessible UI using Github Co-pilot / A Tour of Open Source Foundation Models, Evaluation Frameworks, and Ecosystems
Case Study: Building Accessible UI using Github Co-pilot – Navya Agarwal
Accessibility is more than just a checklist—it’s a fundamental aspect of inclusive web development. However, incorporating AI tools like GitHub Copilot into your workflow presents both new possibilities and unique challenges. In this case study, I’ll share my experience using GitHub Copilot to build accessible web components for Adobe Express. Join me to learn how to leverage AI tools like Copilot effectively, without losing sight of the critical role you play as a developer.
From the excitement of speeding up the coding process to the harsh reality of realizing that Copilot’s suggestions are just the beginning, this talk will take you through the ups and downs of ensuring accessibility. I’ll share the accessibility myths I initially believed, the mistakes I made, and the lessons I learned the hard way. Copilot’s recommendation to use semantic HTML tags instead of generic div elements was a great start, but I quickly discovered that true accessibility goes far beyond what an AI can suggest.
We’ll dissect the Adobe Express website, zooming in on accessibility considerations and the challenges I faced in implementing them. You’ll see how Copilot helped improve interactivity with suggestions like tabindex, ARIA attributes, and keydown event handling—but also why it’s crucial not to take these suggestions at face value. Much like pair programming, Copilot is a powerful tool, but the final responsibility lies with you, the developer.
Join me as we explore compelling use cases for GitHub Copilot, examine its strengths and limitations, and discuss how to navigate complex UI widgets using keyboard events and ARIA roles. By the end of this session, you’ll have a deeper understanding of how to harness AI tools to build not just functional, but truly accessible web applications.
A Tour of Open Source Foundation Models, Evaluation Frameworks, and Ecosystems – Brad Topol
Open Source Foundation Models are emerging as a popular choice for powering AI applications across numerous domains including language, code, time series, and vision. In this talk, we provide an overview of a variety of open source foundation models. Next, we describe recent innovations for evaluating the openness of foundation models and related tooling advances as well. We then present several open source projects that are emerging to form a vast ecosystem for foundation models. Finally, we end with an overview walkthrough of online workshop materials that will help you to gain a deeper understanding of how open source foundation models can be leveraged to address a variety of AI use cases and applications.
