Bots are People Too (Kind Of): Why AI Needs Accessibility

Ever imagined our websites and mobile apps needing to be accessible not just for humans, but for robots too? It sounds like a Black Mirror episode, but with AI Agents becoming increasingly integrated into our digital lives, it’s a reality we need to face. Those AI Agents could be acting as proxies for consumers in their interactions with brands or they could be enterprises’ own agents acting on internal systems. Either way, we need to start rethinking how we do digital design to incorporate the needs of these new AI technologies.

Think of a visually impaired person using a screen reader to navigate a website. They rely on proper headings and alt text to understand the content. Now, imagine an AI shopping assistant trying to do the same thing. It needs to process the entire page at once, figure out the visual hierarchy, and understand which “Add to Cart” button goes with which product (especially when there are multiple options on the screen).

Humans and AI both need accessibility, but they approach it differently. In my example above, the screen reader goes through the content step-by-step, while the AI tries to build a complete map of all the elements and their relationships in the blink of an eye.

Semantic Structure: The Universal Language (for Humans and Robots)

The key to making our websites accessible to both humans and AI is good semantic structure. This means using HTML elements like <header>, <nav>, and <main> to create a clear hierarchy that screen readers can interpret for visually impaired users.

And guess what? This same structure is essential for AI Agents trying to understand and interact with web content. When AI Agents encounter well-structured content with clear headings, properly labeled buttons, and logically organized information, they can easily identify interactive elements, prioritize content, and complete tasks on behalf of users.

Think of it this way: if your website is a well-organized library with clear signs and labels, both humans and robots can find what they need without getting lost. But if it’s a cluttered mess, everyone’s going to struggle.

Standardization: Predictability Makes Everyone Happy

Another crucial aspect is standardization and consistency in design patterns. For humans with cognitive disabilities, consistent placement of navigation menus, search functions, and call-to-action buttons makes the website easier to use. AI Agents, especially visual-based systems, also rely on these consistent patterns to recognize elements across different contexts. When brands maintain consistency in their interfaces, AI Agents can confidently execute tasks without getting confused.

Imagine trying to find the “Checkout” button on an e-commerce site, but it’s labeled differently on every page. Frustrating, right? That’s how AI Agents feel when faced with inconsistent design.

A futuristic digital landscape blending real-world and virtual elements, illustrating AI accessibility. A humanoid robot with glowing blue eyes gestures towards a structured, web-like interface floating over a cityscape. The environment features a road labeled with website sections like 'HEADER,' 'NAVIGATION,' 'CONTENT,' and 'FOOTER,' symbolizing semantic structure. A human figure walks along the digital street, while wireframe buildings, AI icons, and construction elements depict an evolving, AI-friendly digital ecosystem.

The Curb-Cut Effect in the Digital World

You know how curb cuts, those little ramps connecting sidewalks to streets, were originally designed for people with disabilities but ended up benefiting everyone? Strollers, delivery folks, even skateboarders – everyone uses them now. That’s the curb-cut effect in action, and it’s happening in the digital world too. When we design websites and apps with accessibility in mind, we create a better experience for all users, both human and AI.

Think about it: clear headings, descriptive labels, and consistent layouts not only help people with disabilities navigate a website, but they also make it easier for AI Agents to understand and interact with the content. It’s like building a digital curb cut that makes the online world more accessible to everyone.

So, by embracing accessibility, we’re not just doing the right thing, we’re also creating a more user-friendly and inclusive digital environment for everyone, including our new AI companions.

So, What Can You Do?

As we get more experiences with AI Agents interacting with our brands, new design best practices will emerge—but it is clearly still early days. But there are a few steps you can take today to make your digital properties more AI Agent accessible:

  • Implement Comprehensive Structured Data and Metadata: Go beyond basic semantic HTML by using schema.org markup and JSON-LD to provide more context about your content. This helps AI Agents understand your website better and makes it more accessible to assistive technologies.
  • Adopt Accessibility-Driven Testing and Maintenance: Integrate automated accessibility testing into your development process to identify issues affecting both human and AI users. Regular testing and updates will ensure your website remains accessible as content changes and technologies evolve.
  • Anticipate “Search-to-Checkout” Conversations: The most successful use cases will move from conversations to transactions. Accessibility is needed throughout.

Bot Accessibility Applies to All Media and Channels

The above suggestions apply to websites. There are equally important considerations that apply to voice channels, where AI agents are positioned to augment or replace speech-enabled interactive voice response (IVR) systems. A recent viral video depicted how an automated voice assistant that called into an enterprise virtual assistant and conversed using words until they discovered it to be more efficient to handshake using modem tones and converse using computer noises.

Although this “Gibberlink” mode was just the outcome of a hackathon, it suggests that enterprises are ill-prepared to handle the sort of inbound traffic that services like Google’s Duplex/Ask For Me or Gemini are already generating.



Categories: Conversational Intelligence, Intelligent Assistants, Articles

Tags: , , , ,