The rapid advancements in AI are changing how we interact with digital platforms. Among these, agentic AI—intelligent systems capable of making decisions and completing tasks autonomously—is poised to redefine digital accessibility.
AI agents, such as Project Mariner from Google DeepMind’s research, are increasingly adept at analysing website elements to locate information and complete tasks. Unlike humans, these systems evaluate both the underlying code and visual elements of a webpage within milliseconds. For example, an AI agent searching for a button doesn’t just look at its label or functionality—it can consider its position on the page, visual emphasis, and accessibility features like ARIA tags and keyboard navigability.
Accessibility as a Ranking Factor
While AI is set to revolutionally enhance accessibility by assisting users in navigating websites, there is also another vital angle to consider: the accessibility of a website itself may become a ranking factor evaluated by AI agents. For businesses in the finance sector, where trust and inclusivity significantly influence customer decision-making, accessibility may soon become a critical factor evaluated by AI agents on behalf of users.
While human users might overlook accessibility issues such as screen reader compatibility, AI agents will not. They scrutinize every detail, from WCAG (Web Content Accessibility Guidelines) compliance to interactive element usability, and assess whether a company’s digital presence aligns with its values. For instance, a company that claims inclusivity as a core value but has poor contrast ratios and missing alt text will be flagged by AI agents. Such evaluations directly influence the recommendations made to end-users, where trust and inclusivity are paramount.
Accessibility Review: The Big Four
To explore how accessibility compliance varies, I reviewed the homepages of the four largest international accounting firms using Google Chrome’s Lighthouse accessibility tool. Lighthouse generates scores out of 100 by assessing various accessibility metrics, offering a snapshot of a website’s performance against critical accessibility standards.
Here’s how Lighthouse assessed the homepages of The Big Fours' websites, giving an overall accessibility score:
Lighthouse identified the following accessibility issues:
- Headings were found to skip levels, making it harder for assistive technologies to navigate the page, such as screen readers.
- Some text had low contrast, making it difficult or impossible for many users to read, including those with low vision.
- Touch targets were too small or lacked sufficient spacing, making it difficult for users to activate them, especially those with motor impairments.
- Various ARIA tag errors were detected, impacting functionality for assistive technologies.
AI agents, much like Lighthouse, can be trained to assess websites against frameworks like WCAG 2.0 AA—a minimum legal requirement for the UK's public sector. This capability ensures there is no hiding accessibility flaws from AI evaluations. Companies unable to meet these standards risk losing potential customers who prioritise inclusivity in their decision-making process.
Final Thoughts
As AI agents gain prominence, compliance with accessibility standards is no longer just about serving human users. It also means meeting the rigorous expectations of these AI systems, which can evaluate websites with unparalleled precision. Companies that prioritise accessibility today will not only enhance user experiences for all, but also position themselves as leading brands to AI agents, where digital inclusivity will be a worthy mark of success.
The intersection of accessibility and AI also raises important questions about the future evolution of WCAG standards. As AI agents increasingly evaluate websites for accessibility, will the guidelines adapt to reflect this shift? Future updates might introduce new requirements that ensure AI agents can accurately assess and interact with digital content, aligning both human and AI needs.
