{"id":233,"date":"2026-03-02T17:07:38","date_gmt":"2026-03-02T17:07:38","guid":{"rendered":"https:\/\/pressbooks.hcfl.edu\/hccdigitalaccessibility\/?post_type=chapter&#038;p=233"},"modified":"2026-04-12T18:55:14","modified_gmt":"2026-04-12T18:55:14","slug":"agentic-ai-and-digital-accessibilty","status":"publish","type":"chapter","link":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/chapter\/agentic-ai-and-digital-accessibilty\/","title":{"raw":"Agentic Artificial Intelligence and the Future of Digital Accessibility","rendered":"Agentic Artificial Intelligence and the Future of Digital Accessibility"},"content":{"raw":"<div class=\"agentic-ai-and-digital-accessibility-chapter\">\r\n<h2>Agentic Artificial Intelligence and the Future of Digital Accessibility<\/h2>\r\nDigital accessibility has traditionally focused on making websites, documents, and applications usable for people with disabilities through standards such as the Web Content Accessibility Guidelines (WCAG). A major technological shift is now underway. Artificial intelligence systems are evolving from passive tools that respond to commands into agentic systems capable of making decisions, completing multi-step tasks, and acting on behalf of users. Agentic AI may become one of the most significant developments in accessibility since the rise of the internet itself.\r\n\r\nThis chapter introduces agentic artificial intelligence and examines how autonomous AI systems may change digital accessibility. It explores how AI may expand access, personalize interactions, and automate accessibility supports, while also creating new risks related to bias, privacy, accountability, and user control. The chapter encourages readers to think critically about the future of accessibility in AI-driven digital environments.\r\n<div class=\"textbox textbox--learning-objectives\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Learning Objectives<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n\r\nAfter completing this chapter, you will be able to:\r\n<ul>\r\n \t<li>Define agentic artificial intelligence and distinguish it from traditional AI systems.<\/li>\r\n \t<li>Explain how agentic AI technologies may affect digital accessibility.<\/li>\r\n \t<li>Identify opportunities and risks of autonomous AI systems for people with disabilities.<\/li>\r\n \t<li>Evaluate accessibility implications of AI-driven digital environments.<\/li>\r\n \t<li>Describe ethical responsibilities when deploying agentic AI in education and public services.<\/li>\r\n<\/ul>\r\n<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--key-terms\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Key Terms<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<ul>\r\n \t<li><strong>Agentic AI:<\/strong> Artificial intelligence systems that can interpret goals, plan actions, complete tasks, and adapt based on outcomes with limited human direction.<\/li>\r\n \t<li><strong>Assistive Technology:<\/strong> Tools that help individuals with disabilities interact with digital systems, such as screen readers, speech recognition software, and alternative input devices.<\/li>\r\n \t<li><strong>Adaptive Accessibility:<\/strong> Accessibility approaches that dynamically adjust content, interfaces, or interactions to match user needs and preferences.<\/li>\r\n \t<li><strong>Algorithmic Bias:<\/strong> Systematic errors in AI outputs caused by biased data, design choices, or modeling assumptions.<\/li>\r\n \t<li><strong>Human-Centered Design:<\/strong> A design approach that prioritizes human needs, usability, and inclusion throughout the development process.<\/li>\r\n \t<li><strong>Autonomous System:<\/strong> A technology that can carry out actions or make decisions without continuous human direction.<\/li>\r\n<\/ul>\r\n<\/div>\r\n<\/div>\r\n<h3>Short Video: AI and Accessibility<\/h3>\r\nThe following short video explains how artificial intelligence can support accessibility and highlights the importance of responsible design.\r\n\r\nhttps:\/\/www.youtube.com\/watch?v=1Lkfb8MZDBo\r\n\r\n[h5p id=\"16\"]\r\n<div class=\"textbox textbox--accessibility\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Accessibility Note<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">When embedding or linking to video, make sure captions and a transcript are available for learners who need text-based access to audio content.<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--accessibility\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Video Transcript<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n\r\n<strong>Clean transcript:<\/strong> Artificial intelligence can help improve accessibility by supporting tasks such as captioning, image description, communication, and content adaptation. These tools may help reduce barriers for people with disabilities, but they must be designed responsibly. AI systems should be fair, reliable, and inclusive so they support access rather than create new obstacles.\r\n\r\n<\/div>\r\n<\/div>\r\n<h3>Introduction: Accessibility at an AI Turning Point<\/h3>\r\nDigital accessibility has often focused on ensuring that users can access information and complete tasks within systems that may not have been designed with disability access in mind. Agentic AI changes this model. Instead of only providing tools that help users interpret interfaces, AI systems may increasingly act as intermediaries that help users navigate digital environments, translate information, and complete complex tasks.\r\n\r\nThis shift suggests a new accessibility model in which technology adapts to the user more actively. At its best, agentic AI could support greater independence, personalization, and flexibility. At its worst, it could automate exclusion, reduce user control, and create new barriers that are harder to detect.\r\n<h3>What Is Agentic AI?<\/h3>\r\nAgentic AI refers to artificial intelligence systems that can perform goal-directed actions with a degree of autonomy. Unlike traditional software that waits for direct input at each step, agentic systems may be able to:\r\n<ul>\r\n \t<li>Understand goals<\/li>\r\n \t<li>Plan actions<\/li>\r\n \t<li>Execute tasks autonomously<\/li>\r\n \t<li>Learn from outcomes<\/li>\r\n \t<li>Interact across multiple digital environments<\/li>\r\n<\/ul>\r\nExamples may include AI assistants that complete online forms, navigate websites, schedule appointments, gather information, or communicate with other systems on behalf of a user.\r\n<h3>From Assistive Technology to Autonomous Assistance<\/h3>\r\nHistorically, accessibility has depended on specialized assistive technologies that help users work within existing digital systems.\r\n<h4>Traditional Accessibility Model<\/h4>\r\n<ul>\r\n \t<li>Screen readers<\/li>\r\n \t<li>Speech recognition software<\/li>\r\n \t<li>Alternative input devices<\/li>\r\n \t<li>Captioning systems<\/li>\r\n<\/ul>\r\n<h4>Agentic Accessibility Model<\/h4>\r\nAgentic AI introduces a new paradigm in which technology becomes a more active accessibility partner. Instead of only helping users interpret content, AI agents may:\r\n<ul>\r\n \t<li>Interpret inaccessible interfaces<\/li>\r\n \t<li>Convert content formats automatically<\/li>\r\n \t<li>Work around digital barriers in real time<\/li>\r\n \t<li>Personalize interaction methods<\/li>\r\n<\/ul>\r\nFor example, an AI agent might navigate a poorly designed website, extract the needed information, and present it in a simpler and more accessible format.\r\n<div class=\"textbox textbox--tip\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Practical Tip<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">When discussing agentic AI with students or staff, compare it with traditional assistive technology so learners can see how autonomy changes the accessibility model.<\/div>\r\n<\/div>\r\n<h3>Opportunities for Digital Accessibility<\/h3>\r\n<h4>Personalized Accessibility<\/h4>\r\nAI systems may adapt to individual needs and preferences rather than offering the same accessibility support to every user. Personalization may include:\r\n<ul>\r\n \t<li>Reading complexity levels<\/li>\r\n \t<li>Preferred interaction methods<\/li>\r\n \t<li>Cognitive load tolerance<\/li>\r\n \t<li>Sensory sensitivities<\/li>\r\n<\/ul>\r\nThis could shift accessibility from standardized compliance toward more adaptive support.\r\n<h4>Real-Time Accessibility Remediation<\/h4>\r\nAgentic systems may automatically improve inaccessible content by:\r\n<ul>\r\n \t<li>Generating image descriptions<\/li>\r\n \t<li>Repairing inaccessible PDFs<\/li>\r\n \t<li>Creating captions and transcripts<\/li>\r\n \t<li>Simplifying complex language<\/li>\r\n \t<li>Reformatting layouts for readability<\/li>\r\n<\/ul>\r\nThese capabilities may reduce the need for manual retrofitting, although they should not replace accessible design from the start.\r\n<h4>Expanded Independence<\/h4>\r\nAI agents may help users independently manage tasks that are often difficult in inaccessible digital environments, such as:\r\n<ul>\r\n \t<li>Completing bureaucratic processes<\/li>\r\n \t<li>Managing educational platforms<\/li>\r\n \t<li>Using employment systems<\/li>\r\n \t<li>Navigating healthcare portals<\/li>\r\n<\/ul>\r\nThis represents a shift from accommodation toward greater digital autonomy.\r\n<h4>Accessibility in Education<\/h4>\r\nIn higher education, agentic AI may support:\r\n<ul>\r\n \t<li>Adaptive learning environments<\/li>\r\n \t<li>Automatically accessible course materials<\/li>\r\n \t<li>Intelligent tutoring systems<\/li>\r\n \t<li>Real-time lecture accessibility supports<\/li>\r\n<\/ul>\r\nStudents may increasingly interact with course systems through personalized AI intermediaries that help them access content and complete academic tasks.\r\n<h3>Risks and Accessibility Challenges<\/h3>\r\nDespite its promise, agentic AI introduces new risks that educators, institutions, and designers must consider carefully.\r\n<h4>Algorithmic Bias<\/h4>\r\nAI systems trained on incomplete or biased datasets may:\r\n<ul>\r\n \t<li>Misinterpret disability-related behaviors<\/li>\r\n \t<li>Produce inaccurate captions or descriptions<\/li>\r\n \t<li>Exclude marginalized users<\/li>\r\n<\/ul>\r\nAccessibility failures may become automated and scaled across systems.\r\n<h4>Loss of User Control<\/h4>\r\nAutonomous systems raise important questions about control and accountability. For example:\r\n<ul>\r\n \t<li>Who controls accessibility preferences?<\/li>\r\n \t<li>Can users override AI decisions?<\/li>\r\n \t<li>What happens when an AI system acts incorrectly?<\/li>\r\n<\/ul>\r\nAccessibility should empower users, not replace their agency.\r\n<h4>Over-Reliance on AI Fixes<\/h4>\r\nOrganizations may be tempted to rely on AI overlays or automated remediation instead of building accessible systems from the start. This creates a risk that accessibility becomes reactive rather than foundational. Accessible design must remain a core responsibility.\r\n<h4>Privacy and Surveillance Concerns<\/h4>\r\nAgentic accessibility systems may require sensitive personal information, including:\r\n<ul>\r\n \t<li>Disability-related information<\/li>\r\n \t<li>Behavioral patterns<\/li>\r\n \t<li>Communication styles<\/li>\r\n \t<li>Personal preferences<\/li>\r\n<\/ul>\r\nEthical implementation requires clear consent practices, strong privacy protections, and transparent data governance.\r\n<div class=\"textbox textbox--warning\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Important Note<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">AI systems should not remove user choice. People must be able to review, override, or reject automated accessibility decisions when needed.<\/div>\r\n<\/div>\r\n<h3>Rethinking Accessibility Frameworks<\/h3>\r\nCurrent standards such as WCAG are primarily designed to evaluate interfaces, content, and interactions that can be inspected in relatively stable ways. Agentic environments introduce new questions that are harder to measure with a static checklist.\r\n\r\nFuture accessibility frameworks may need to consider:\r\n<ul>\r\n \t<li>Decision transparency<\/li>\r\n \t<li>User override capability<\/li>\r\n \t<li>Fairness in personalization<\/li>\r\n \t<li>Continuous accessibility performance<\/li>\r\n<\/ul>\r\nIn AI-driven environments, accessibility may become an ongoing process of monitoring, testing, and revision rather than a one-time compliance task.\r\n<div class=\"textbox textbox--accessibility\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Accessibility Check<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<ul>\r\n \t<li>Can users understand how the AI system made a decision?<\/li>\r\n \t<li>Can users override automated choices?<\/li>\r\n \t<li>Does personalization remain fair across users?<\/li>\r\n \t<li>Is accessibility monitored over time rather than checked only once?<\/li>\r\n<\/ul>\r\n<\/div>\r\n<\/div>\r\n<h3>Human-AI Collaboration and Inclusive Design<\/h3>\r\nThe future of accessibility is unlikely to be fully automated. Strong accessibility practice will still require:\r\n<ul>\r\n \t<li>Human-centered design<\/li>\r\n \t<li>Input from disability communities<\/li>\r\n \t<li>Ethical AI governance<\/li>\r\n \t<li>Continuous evaluation<\/li>\r\n<\/ul>\r\nPeople with disabilities should remain active participants in shaping agentic systems, not simply end users who experience the consequences of design decisions made by others.\r\n<h3>The Role of Higher Education<\/h3>\r\nColleges and universities play an important role in shaping accessible AI futures. Institutions can:\r\n<ul>\r\n \t<li>Teach responsible AI development<\/li>\r\n \t<li>Model accessible AI adoption<\/li>\r\n \t<li>Evaluate emerging technologies critically<\/li>\r\n \t<li>Prepare students for inclusive digital workplaces<\/li>\r\n<\/ul>\r\nDigital accessibility education now increasingly intersects with AI literacy, ethics, and public responsibility.\r\n<div class=\"textbox textbox--summary\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Chapter Summary<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">Agentic AI represents a significant shift in digital accessibility. These systems may support unprecedented personalization, automation, and independence for users with disabilities. At the same time, they introduce serious challenges related to bias, privacy, accountability, and over-reliance on automated fixes. The future of accessibility will depend not only on technical innovation, but also on inclusive design, ethical governance, and ongoing collaboration with disability communities.<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--key-takeaways\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Key Takeaways<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<ul>\r\n \t<li>Agentic AI systems can pursue goals and complete tasks with limited human direction.<\/li>\r\n \t<li>These systems may improve digital accessibility through personalization, automation, and adaptive support.<\/li>\r\n \t<li>Agentic AI also introduces risks related to bias, privacy, accountability, and loss of user control.<\/li>\r\n \t<li>Accessible design must remain foundational even when AI tools are available.<\/li>\r\n \t<li>Higher education institutions have a responsibility to teach, evaluate, and model ethical accessibility practices in AI-driven environments.<\/li>\r\n<\/ul>\r\n<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--exercises\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Review Questions<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<ol>\r\n \t<li>How does agentic AI differ from traditional AI systems?<\/li>\r\n \t<li>What new accessibility opportunities might autonomous AI systems create?<\/li>\r\n \t<li>What risks emerge when accessibility decisions are automated?<\/li>\r\n \t<li>Why is user control important in AI-supported accessibility systems?<\/li>\r\n \t<li>How might colleges and universities help shape more accessible AI futures?<\/li>\r\n<\/ol>\r\n<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--exercises\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Applied Activity<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<h4>Accessibility Futures Analysis<\/h4>\r\nChoose a digital system used at your institution, such as a learning management system, registration portal, or website.\r\n\r\nEvaluate the following:\r\n<ul>\r\n \t<li>How an AI agent could improve accessibility<\/li>\r\n \t<li>Potential risks introduced by automation<\/li>\r\n \t<li>What human oversight would still be required<\/li>\r\n<\/ul>\r\nPrepare a short accessibility impact report that explains both the promise and the risks of agentic AI in that environment.\r\n\r\n<\/div>\r\n<\/div>\r\n<div class=\"textbox textbox--exercises\"><header class=\"textbox__header\">\r\n<h3 class=\"textbox__title\">Further Reading<\/h3>\r\n<\/header>\r\n<div class=\"textbox__content\">\r\n<ul>\r\n \t<li><a href=\"https:\/\/www.microsoft.com\/en-us\/accessibility\/resources\">Explore Microsoft accessibility resources<\/a><\/li>\r\n \t<li><a href=\"https:\/\/www.w3.org\/WAI\/\">Visit the W3C Web Accessibility Initiative<\/a><\/li>\r\n \t<li><a href=\"https:\/\/www.w3.org\/WAI\/standards-guidelines\/wcag\/\">Read the Web Content Accessibility Guidelines<\/a><\/li>\r\n \t<li><a href=\"https:\/\/www.section508.gov\/\">Learn about Section 508 requirements<\/a><\/li>\r\n<\/ul>\r\n<\/div>\r\n<\/div>\r\n<\/div>\r\n<div class=\"textbox\">\r\n<h3><strong>Licenses and Attribution<\/strong><\/h3>\r\n<h4>CC Licensed Content, Original<\/h4>\r\nThis educational material includes AI-generated content from ChatGPT by OpenAI. The original content created by Josh Hill, Neida Abraham, and Emiliana Olavarrieta from Hillsborough College is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (<a href=\"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/deed.en\" rel=\"noreferrer noopener\">CC BY-NC 4.0<\/a>).\r\n\r\nAll images in this textbook generated with DALL\u00b7E are licensed under the terms provided by OpenAI, allowing their use, modification, and distribution with appropriate attribution.\r\n<h4>Other Licensed Content<\/h4>\r\n<strong>AI and Accessibility<\/strong>\r\nMSFTEnable\r\nLicense: Standard YouTube License.\r\n\r\n<\/div>","rendered":"<div class=\"agentic-ai-and-digital-accessibility-chapter\">\n<h2>Agentic Artificial Intelligence and the Future of Digital Accessibility<\/h2>\n<p>Digital accessibility has traditionally focused on making websites, documents, and applications usable for people with disabilities through standards such as the Web Content Accessibility Guidelines (WCAG). A major technological shift is now underway. Artificial intelligence systems are evolving from passive tools that respond to commands into agentic systems capable of making decisions, completing multi-step tasks, and acting on behalf of users. Agentic AI may become one of the most significant developments in accessibility since the rise of the internet itself.<\/p>\n<p>This chapter introduces agentic artificial intelligence and examines how autonomous AI systems may change digital accessibility. It explores how AI may expand access, personalize interactions, and automate accessibility supports, while also creating new risks related to bias, privacy, accountability, and user control. The chapter encourages readers to think critically about the future of accessibility in AI-driven digital environments.<\/p>\n<div class=\"textbox textbox--learning-objectives\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Learning Objectives<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<p>After completing this chapter, you will be able to:<\/p>\n<ul>\n<li>Define agentic artificial intelligence and distinguish it from traditional AI systems.<\/li>\n<li>Explain how agentic AI technologies may affect digital accessibility.<\/li>\n<li>Identify opportunities and risks of autonomous AI systems for people with disabilities.<\/li>\n<li>Evaluate accessibility implications of AI-driven digital environments.<\/li>\n<li>Describe ethical responsibilities when deploying agentic AI in education and public services.<\/li>\n<\/ul>\n<\/div>\n<\/div>\n<div class=\"textbox textbox--key-terms\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Key Terms<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<ul>\n<li><strong>Agentic AI:<\/strong> Artificial intelligence systems that can interpret goals, plan actions, complete tasks, and adapt based on outcomes with limited human direction.<\/li>\n<li><strong>Assistive Technology:<\/strong> Tools that help individuals with disabilities interact with digital systems, such as screen readers, speech recognition software, and alternative input devices.<\/li>\n<li><strong>Adaptive Accessibility:<\/strong> Accessibility approaches that dynamically adjust content, interfaces, or interactions to match user needs and preferences.<\/li>\n<li><strong>Algorithmic Bias:<\/strong> Systematic errors in AI outputs caused by biased data, design choices, or modeling assumptions.<\/li>\n<li><strong>Human-Centered Design:<\/strong> A design approach that prioritizes human needs, usability, and inclusion throughout the development process.<\/li>\n<li><strong>Autonomous System:<\/strong> A technology that can carry out actions or make decisions without continuous human direction.<\/li>\n<\/ul>\n<\/div>\n<\/div>\n<h3>Short Video: AI and Accessibility<\/h3>\n<p>The following short video explains how artificial intelligence can support accessibility and highlights the importance of responsible design.<\/p>\n<p><iframe loading=\"lazy\" id=\"oembed-1\" title=\"AI and accessibility\" width=\"500\" height=\"281\" src=\"https:\/\/www.youtube.com\/embed\/1Lkfb8MZDBo?feature=oembed&#38;rel=0\" frameborder=\"0\" allowfullscreen=\"allowfullscreen\"><\/iframe><\/p>\n<div id=\"h5p-16\">\n<div class=\"h5p-iframe-wrapper\"><iframe id=\"h5p-iframe-16\" class=\"h5p-iframe\" data-content-id=\"16\" style=\"height:1px\" src=\"about:blank\" frameBorder=\"0\" scrolling=\"no\" title=\"Agentic AI and future of Digital accessibility\"><\/iframe><\/div>\n<\/div>\n<div class=\"textbox textbox--accessibility\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Accessibility Note<\/h3>\n<\/header>\n<div class=\"textbox__content\">When embedding or linking to video, make sure captions and a transcript are available for learners who need text-based access to audio content.<\/div>\n<\/div>\n<div class=\"textbox textbox--accessibility\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Video Transcript<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<p><strong>Clean transcript:<\/strong> Artificial intelligence can help improve accessibility by supporting tasks such as captioning, image description, communication, and content adaptation. These tools may help reduce barriers for people with disabilities, but they must be designed responsibly. AI systems should be fair, reliable, and inclusive so they support access rather than create new obstacles.<\/p>\n<\/div>\n<\/div>\n<h3>Introduction: Accessibility at an AI Turning Point<\/h3>\n<p>Digital accessibility has often focused on ensuring that users can access information and complete tasks within systems that may not have been designed with disability access in mind. Agentic AI changes this model. Instead of only providing tools that help users interpret interfaces, AI systems may increasingly act as intermediaries that help users navigate digital environments, translate information, and complete complex tasks.<\/p>\n<p>This shift suggests a new accessibility model in which technology adapts to the user more actively. At its best, agentic AI could support greater independence, personalization, and flexibility. At its worst, it could automate exclusion, reduce user control, and create new barriers that are harder to detect.<\/p>\n<h3>What Is Agentic AI?<\/h3>\n<p>Agentic AI refers to artificial intelligence systems that can perform goal-directed actions with a degree of autonomy. Unlike traditional software that waits for direct input at each step, agentic systems may be able to:<\/p>\n<ul>\n<li>Understand goals<\/li>\n<li>Plan actions<\/li>\n<li>Execute tasks autonomously<\/li>\n<li>Learn from outcomes<\/li>\n<li>Interact across multiple digital environments<\/li>\n<\/ul>\n<p>Examples may include AI assistants that complete online forms, navigate websites, schedule appointments, gather information, or communicate with other systems on behalf of a user.<\/p>\n<h3>From Assistive Technology to Autonomous Assistance<\/h3>\n<p>Historically, accessibility has depended on specialized assistive technologies that help users work within existing digital systems.<\/p>\n<h4>Traditional Accessibility Model<\/h4>\n<ul>\n<li>Screen readers<\/li>\n<li>Speech recognition software<\/li>\n<li>Alternative input devices<\/li>\n<li>Captioning systems<\/li>\n<\/ul>\n<h4>Agentic Accessibility Model<\/h4>\n<p>Agentic AI introduces a new paradigm in which technology becomes a more active accessibility partner. Instead of only helping users interpret content, AI agents may:<\/p>\n<ul>\n<li>Interpret inaccessible interfaces<\/li>\n<li>Convert content formats automatically<\/li>\n<li>Work around digital barriers in real time<\/li>\n<li>Personalize interaction methods<\/li>\n<\/ul>\n<p>For example, an AI agent might navigate a poorly designed website, extract the needed information, and present it in a simpler and more accessible format.<\/p>\n<div class=\"textbox textbox--tip\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Practical Tip<\/h3>\n<\/header>\n<div class=\"textbox__content\">When discussing agentic AI with students or staff, compare it with traditional assistive technology so learners can see how autonomy changes the accessibility model.<\/div>\n<\/div>\n<h3>Opportunities for Digital Accessibility<\/h3>\n<h4>Personalized Accessibility<\/h4>\n<p>AI systems may adapt to individual needs and preferences rather than offering the same accessibility support to every user. Personalization may include:<\/p>\n<ul>\n<li>Reading complexity levels<\/li>\n<li>Preferred interaction methods<\/li>\n<li>Cognitive load tolerance<\/li>\n<li>Sensory sensitivities<\/li>\n<\/ul>\n<p>This could shift accessibility from standardized compliance toward more adaptive support.<\/p>\n<h4>Real-Time Accessibility Remediation<\/h4>\n<p>Agentic systems may automatically improve inaccessible content by:<\/p>\n<ul>\n<li>Generating image descriptions<\/li>\n<li>Repairing inaccessible PDFs<\/li>\n<li>Creating captions and transcripts<\/li>\n<li>Simplifying complex language<\/li>\n<li>Reformatting layouts for readability<\/li>\n<\/ul>\n<p>These capabilities may reduce the need for manual retrofitting, although they should not replace accessible design from the start.<\/p>\n<h4>Expanded Independence<\/h4>\n<p>AI agents may help users independently manage tasks that are often difficult in inaccessible digital environments, such as:<\/p>\n<ul>\n<li>Completing bureaucratic processes<\/li>\n<li>Managing educational platforms<\/li>\n<li>Using employment systems<\/li>\n<li>Navigating healthcare portals<\/li>\n<\/ul>\n<p>This represents a shift from accommodation toward greater digital autonomy.<\/p>\n<h4>Accessibility in Education<\/h4>\n<p>In higher education, agentic AI may support:<\/p>\n<ul>\n<li>Adaptive learning environments<\/li>\n<li>Automatically accessible course materials<\/li>\n<li>Intelligent tutoring systems<\/li>\n<li>Real-time lecture accessibility supports<\/li>\n<\/ul>\n<p>Students may increasingly interact with course systems through personalized AI intermediaries that help them access content and complete academic tasks.<\/p>\n<h3>Risks and Accessibility Challenges<\/h3>\n<p>Despite its promise, agentic AI introduces new risks that educators, institutions, and designers must consider carefully.<\/p>\n<h4>Algorithmic Bias<\/h4>\n<p>AI systems trained on incomplete or biased datasets may:<\/p>\n<ul>\n<li>Misinterpret disability-related behaviors<\/li>\n<li>Produce inaccurate captions or descriptions<\/li>\n<li>Exclude marginalized users<\/li>\n<\/ul>\n<p>Accessibility failures may become automated and scaled across systems.<\/p>\n<h4>Loss of User Control<\/h4>\n<p>Autonomous systems raise important questions about control and accountability. For example:<\/p>\n<ul>\n<li>Who controls accessibility preferences?<\/li>\n<li>Can users override AI decisions?<\/li>\n<li>What happens when an AI system acts incorrectly?<\/li>\n<\/ul>\n<p>Accessibility should empower users, not replace their agency.<\/p>\n<h4>Over-Reliance on AI Fixes<\/h4>\n<p>Organizations may be tempted to rely on AI overlays or automated remediation instead of building accessible systems from the start. This creates a risk that accessibility becomes reactive rather than foundational. Accessible design must remain a core responsibility.<\/p>\n<h4>Privacy and Surveillance Concerns<\/h4>\n<p>Agentic accessibility systems may require sensitive personal information, including:<\/p>\n<ul>\n<li>Disability-related information<\/li>\n<li>Behavioral patterns<\/li>\n<li>Communication styles<\/li>\n<li>Personal preferences<\/li>\n<\/ul>\n<p>Ethical implementation requires clear consent practices, strong privacy protections, and transparent data governance.<\/p>\n<div class=\"textbox textbox--warning\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Important Note<\/h3>\n<\/header>\n<div class=\"textbox__content\">AI systems should not remove user choice. People must be able to review, override, or reject automated accessibility decisions when needed.<\/div>\n<\/div>\n<h3>Rethinking Accessibility Frameworks<\/h3>\n<p>Current standards such as WCAG are primarily designed to evaluate interfaces, content, and interactions that can be inspected in relatively stable ways. Agentic environments introduce new questions that are harder to measure with a static checklist.<\/p>\n<p>Future accessibility frameworks may need to consider:<\/p>\n<ul>\n<li>Decision transparency<\/li>\n<li>User override capability<\/li>\n<li>Fairness in personalization<\/li>\n<li>Continuous accessibility performance<\/li>\n<\/ul>\n<p>In AI-driven environments, accessibility may become an ongoing process of monitoring, testing, and revision rather than a one-time compliance task.<\/p>\n<div class=\"textbox textbox--accessibility\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Accessibility Check<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<ul>\n<li>Can users understand how the AI system made a decision?<\/li>\n<li>Can users override automated choices?<\/li>\n<li>Does personalization remain fair across users?<\/li>\n<li>Is accessibility monitored over time rather than checked only once?<\/li>\n<\/ul>\n<\/div>\n<\/div>\n<h3>Human-AI Collaboration and Inclusive Design<\/h3>\n<p>The future of accessibility is unlikely to be fully automated. Strong accessibility practice will still require:<\/p>\n<ul>\n<li>Human-centered design<\/li>\n<li>Input from disability communities<\/li>\n<li>Ethical AI governance<\/li>\n<li>Continuous evaluation<\/li>\n<\/ul>\n<p>People with disabilities should remain active participants in shaping agentic systems, not simply end users who experience the consequences of design decisions made by others.<\/p>\n<h3>The Role of Higher Education<\/h3>\n<p>Colleges and universities play an important role in shaping accessible AI futures. Institutions can:<\/p>\n<ul>\n<li>Teach responsible AI development<\/li>\n<li>Model accessible AI adoption<\/li>\n<li>Evaluate emerging technologies critically<\/li>\n<li>Prepare students for inclusive digital workplaces<\/li>\n<\/ul>\n<p>Digital accessibility education now increasingly intersects with AI literacy, ethics, and public responsibility.<\/p>\n<div class=\"textbox textbox--summary\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Chapter Summary<\/h3>\n<\/header>\n<div class=\"textbox__content\">Agentic AI represents a significant shift in digital accessibility. These systems may support unprecedented personalization, automation, and independence for users with disabilities. At the same time, they introduce serious challenges related to bias, privacy, accountability, and over-reliance on automated fixes. The future of accessibility will depend not only on technical innovation, but also on inclusive design, ethical governance, and ongoing collaboration with disability communities.<\/div>\n<\/div>\n<div class=\"textbox textbox--key-takeaways\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Key Takeaways<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<ul>\n<li>Agentic AI systems can pursue goals and complete tasks with limited human direction.<\/li>\n<li>These systems may improve digital accessibility through personalization, automation, and adaptive support.<\/li>\n<li>Agentic AI also introduces risks related to bias, privacy, accountability, and loss of user control.<\/li>\n<li>Accessible design must remain foundational even when AI tools are available.<\/li>\n<li>Higher education institutions have a responsibility to teach, evaluate, and model ethical accessibility practices in AI-driven environments.<\/li>\n<\/ul>\n<\/div>\n<\/div>\n<div class=\"textbox textbox--exercises\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Review Questions<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<ol>\n<li>How does agentic AI differ from traditional AI systems?<\/li>\n<li>What new accessibility opportunities might autonomous AI systems create?<\/li>\n<li>What risks emerge when accessibility decisions are automated?<\/li>\n<li>Why is user control important in AI-supported accessibility systems?<\/li>\n<li>How might colleges and universities help shape more accessible AI futures?<\/li>\n<\/ol>\n<\/div>\n<\/div>\n<div class=\"textbox textbox--exercises\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Applied Activity<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<h4>Accessibility Futures Analysis<\/h4>\n<p>Choose a digital system used at your institution, such as a learning management system, registration portal, or website.<\/p>\n<p>Evaluate the following:<\/p>\n<ul>\n<li>How an AI agent could improve accessibility<\/li>\n<li>Potential risks introduced by automation<\/li>\n<li>What human oversight would still be required<\/li>\n<\/ul>\n<p>Prepare a short accessibility impact report that explains both the promise and the risks of agentic AI in that environment.<\/p>\n<\/div>\n<\/div>\n<div class=\"textbox textbox--exercises\">\n<header class=\"textbox__header\">\n<h3 class=\"textbox__title\">Further Reading<\/h3>\n<\/header>\n<div class=\"textbox__content\">\n<ul>\n<li><a href=\"https:\/\/www.microsoft.com\/en-us\/accessibility\/resources\">Explore Microsoft accessibility resources<\/a><\/li>\n<li><a href=\"https:\/\/www.w3.org\/WAI\/\">Visit the W3C Web Accessibility Initiative<\/a><\/li>\n<li><a href=\"https:\/\/www.w3.org\/WAI\/standards-guidelines\/wcag\/\">Read the Web Content Accessibility Guidelines<\/a><\/li>\n<li><a href=\"https:\/\/www.section508.gov\/\">Learn about Section 508 requirements<\/a><\/li>\n<\/ul>\n<\/div>\n<\/div>\n<\/div>\n<div class=\"textbox\">\n<h3><strong>Licenses and Attribution<\/strong><\/h3>\n<h4>CC Licensed Content, Original<\/h4>\n<p>This educational material includes AI-generated content from ChatGPT by OpenAI. The original content created by Josh Hill, Neida Abraham, and Emiliana Olavarrieta from Hillsborough College is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (<a href=\"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/deed.en\" rel=\"noreferrer noopener\">CC BY-NC 4.0<\/a>).<\/p>\n<p>All images in this textbook generated with DALL\u00b7E are licensed under the terms provided by OpenAI, allowing their use, modification, and distribution with appropriate attribution.<\/p>\n<h4>Other Licensed Content<\/h4>\n<p><strong>AI and Accessibility<\/strong><br \/>\nMSFTEnable<br \/>\nLicense: Standard YouTube License.<\/p>\n<\/div>\n","protected":false},"author":2,"menu_order":14,"template":"","meta":{"pb_show_title":"on","pb_short_title":"","pb_subtitle":"","pb_authors":[],"pb_section_license":""},"chapter-type":[],"contributor":[],"license":[],"class_list":["post-233","chapter","type-chapter","status-publish","hentry"],"part":240,"_links":{"self":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapters\/233","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapters"}],"about":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/wp\/v2\/types\/chapter"}],"author":[{"embeddable":true,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/wp\/v2\/users\/2"}],"version-history":[{"count":19,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapters\/233\/revisions"}],"predecessor-version":[{"id":1183,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapters\/233\/revisions\/1183"}],"part":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/parts\/240"}],"metadata":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapters\/233\/metadata\/"}],"wp:attachment":[{"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/wp\/v2\/media?parent=233"}],"wp:term":[{"taxonomy":"chapter-type","embeddable":true,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/pressbooks\/v2\/chapter-type?post=233"},{"taxonomy":"contributor","embeddable":true,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/wp\/v2\/contributor?post=233"},{"taxonomy":"license","embeddable":true,"href":"https:\/\/pressbooks.hcfl.edu\/digitalaccessibility\/wp-json\/wp\/v2\/license?post=233"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}