Generative AI Explored as Tool for Inclusivity
A presentation from AWS's Public Sector team explores how generative AI can be used to improve the inclusivity of digital services. Potential applications include automatically generating accessible content such as image alt text, video captions, and plain language summaries. Experts note that while AI can accelerate accessibility improvements, it is not a substitute for inclusive design principles and thorough user testing.
- The European Union's Web Accessibility Directive (WAD) mandates that public sector websites and mobile apps meet specific accessibility standards, detailed in the harmonized European standard EN 301 549 v3.2.1. This legal framework requires features like accessibility statements and feedback mechanisms for users to report issues. Regular monitoring and reporting by member states ensure ongoing compliance. - Recent updates to the Web Content Accessibility Guidelines (WCAG 2.2), released in October 2023, focus on improving usability for individuals with cognitive or learning disabilities, those with low vision, and users of mobile devices. These changes introduce new success criteria and refine existing ones to better address modern web usage patterns and technologies. - In practice, European public sector bodies are already using AI to improve services; for instance, Spain's Ministry of Justice uses AI for legal document summarization, and Helsinki's "UrbanistAI" platform engages citizens in urban planning. A 2024 report from the European Commission's Joint Research Centre found that 30% of public managers in the EU are already using generative AI for tasks like drafting and summarizing information. - While AI tools can automate accessibility checks for issues like color contrast and missing alt text, they are not a complete replacement for human expertise and user testing. Over-reliance on automated "overlays" can sometimes interfere with users' existing assistive technologies and may not address the root cause of accessibility problems. - The EU's AI Act, finalized in spring 2024 with obligations phasing in through 2026, establishes a regulatory foundation for the use of AI, including generative models. It mandates transparency, ethical compliance, and risk assessments for AI systems used in the public sector. To support this, a new European AI Office was established in June 2024 to oversee governance and ensure compliance. - Several AI-powered tools are being integrated into design and development workflows to identify accessibility issues early. For example, tools like Stark plug into design platforms such as Figma to check for potential issues during the design phase, while others like Axe DevTools integrate into the development process to automate testing. - Challenges in using AI for accessibility include the potential for biased algorithms, which can lead to poor performance for underrepresented groups, such as users with non-standard accents in speech-to-text applications. There are also significant concerns regarding data privacy and the security of sensitive user information collected by AI-driven accessibility software. - Looking ahead, the "GenAI4EU" initiative aims to accelerate the adoption of generative AI across strategic industries and public services in the EU. This includes pilot projects that will explore the use of European GenAI solutions to improve citizen-centric services and internal administrative processes.
