In matters of AI, the future of productivity is being coded today, but not everyone has the access key.
Generative AI tools promise to make work faster, smarter, and more efficient and corporations, for most part, are quick to tout their fiscal benefits, from reduced labour costs to boosted productivity. Yet, it seems that these systems are still designed with a neurotypical, digitally fluent user in mind, leaving those with cognitive, motor, or sensory differences struggling to keep pace with a technology that is already reshaping the workplace around them.
Accessibility in AI vs. Accessibility of AI
In our earlier piece, Is AI Failing Accessibility?, we explored how AI solutions often focus narrowly on certain disabilities, especially visual impairments, while others are overlooked in its development. That diagnosis still stands. But then comes another layer: even when AI tools are designed to enable accessibility, their own interfaces can present new barriers.
And this is where the divide sharpens in the UK. According to Age UK, over in the 65+ age group, more than a third (about 4.7 million people) struggle with even the basics of going online, and another 2.3 million don’t use the internet whatsoever. Pair this with the fact that England and Wales’ population is ageing at pace, and you have a landscape where the very groups who could most benefit from accessible AI are also those least equipped to adopt it.
The irony deepens when viewed globally. The World Health Organisation and the UNICEF estimate that over 2.5 billion people worldwide require assistive technology but nearly 90% lack access to it.
In Europe, lawmakers are beginning to respond to the growing inequity. The EU’s AI Act has placed accessibility at its core, making it a design requirement rather than an afterthought. The UK, by contrast, has yet to define such obligations, risking a scenario where its older and disabled citizens fall even further behind.
The Design Dilemma: Who Are AI Tools Really For?
This beckons a critical question: is designing for vulnerable groups genuinely a priority for the major players shaping the AI landscape? For all the talk of inclusivity, much of the development pipeline in generative AI still leans toward optimising for scale, speed, and profitability, and not so much about levelling the playing field. Accessibility often enters the conversation too late; often as a feature request rather than a founding principle.
From a user experience perspective, this oversight is ethically questionable and short-sighted. Designing for the margins almost always yields solutions that benefit the mainstream. Consider how captions, originally developed for people with hearing impairments, are now indispensable in everyday video consumption — on muted TikToks, in noisy commutes, or for multilingual audiences. The same principle, in practice, applies to AI: interfaces built with neurodiverse, elderly, or disabled users in mind can enhance usability and efficiency for everyone.
Corroborating this, a study published in July 2025 in MDPI’s AI journal by researchers from the University of Cambridge and independent scholars introduces the IDEAS framework, which stands for Inclusion, Diversity, Equity, Accessibility, and Safety—as essential principles for human-centric AI design, emphasizing their integration from the outset of the development process.
The path forward, then, is not simply about making AI available, but making it accessible for everyone. This evolution demands a cultural shift in how we design, regulate, and deploy generative AI. If policymakers embed accessibility as a directive, and if companies adopt frameworks like IDEAS from the outset, the promise of AI could extend beyond efficiency and profit to something far more transformative: a technology that genuinely levels the playing field. The future of generative AI will be judged not by how quickly it scales, but by how inclusively it serves.
So, the real question is this: will AI be remembered as the tool that widened the digital divide, or the one that finally closed it?
