Artificial intelligence (AI) is transforming almost every industry and facet of modern life – enabling new possibilities and opportunities for people with diverse backgrounds, abilities, and needs. Large language models (LLMs) are just one type of AI that can generate natural language texts based on a given input such as a question or keyword. LLMs have shown promising results in various tasks, such as summarization, translation, dialogue, and content creation. Common concerns and challenges, such as data quality, bias, fairness, privacy, and security, are also present with LLMs and other AI models. This session examines the rapidly changing current state and the fragmenting future directions of LLMs with an accessibility and accommodations lens, highlighting some recent cases and innovations, as well as learnings and best practices focused on the full inclusion of people with disabilities.
Learning Objectives:
- Develop a working knowledge of AI and LLMs, including their potential and limitations, and how they can complement and augment existing accessibility and accommodations tools and solutions.
- Gain insights and tips on how to design, develop, and deploy LLMs for accessibility in an ethical and inclusive manner, and how to avoid or mitigate common pitfalls and risks.
- Discuss the potential adoption and application of ethical and inclusive principles and guidelines when creating and using AI and LLMs for accessibility.
Moderator:
- Annette Rosta – Accessibility Leader at KPMG
Panelists:
- Samantha Estoesta – Project Manager, Social Innovation at TD Bank
- Jenny Lay-Flurrie – Vice President, Chief Accessibility Officer at Microsoft
- David Sloan – Chief Accessibility Officer – Vispero
- Thomas Wlodkowski – Vice President, Accessibility at Comcast NBC Universal