How Artificial Intelligence Can Advance Disability Inclusion

Listen to Post

Disclaimer: This article is for general information and awareness purposes only and does not constitute legal, medical, financial, technological or professional advice. Artificial intelligence is developing rapidly, and its capabilities, limitations, regulatory requirements and accessibility features may change over time. Organisations implementing AI should consider applicable equality, accessibility, data-protection and human-rights obligations and seek appropriate professional advice where necessary.

AI and Disability Inclusion: Building an Accessible Future

Artificial intelligence is changing the way people work, communicate, study, access services and interact with the world around them. For persons with disabilities, AI has the potential to be much more than a productivity tool. Used responsibly, it can provide independence, improve communication, reduce barriers and create opportunities that may previously have been inaccessible.

However, technology is not automatically inclusive simply because it is innovative.

If AI systems are designed without considering disability, accessibility and human diversity, they can reproduce existing barriers or create entirely new ones. Facial-recognition systems may struggle to recognise people with facial differences. Automated recruitment systems may disadvantage candidates whose communication styles or employment histories differ from assumed norms. Voice-controlled software may fail to understand people with speech impairments. Websites powered by AI can still be inaccessible to screen-reader users.

An accessible future therefore requires more than advanced technology. It requires inclusive design from the very beginning.

AI as a Tool for Independence

One of the greatest opportunities offered by artificial intelligence is the ability to help people carry out everyday activities more independently.

AI-powered tools can assist with tasks such as:

• Converting speech into written text;
• Converting written information into speech;
• Describing photographs and surroundings;
• Identifying objects;
• Reading documents aloud;
• Summarising complicated information;
• Providing real-time captions;
• Translating languages;
• Helping users organise appointments and reminders;
• Assisting with navigation;
• Predicting text for people who experience difficulties typing;
• Helping people communicate using alternative and augmentative communication systems.

For someone with a visual impairment, AI may help describe an image, recognise an object or read printed text.

For someone who is Deaf or hard of hearing, automated transcription and live captioning may make meetings, lectures and online conversations more accessible.

For people with limited mobility, AI-controlled smart-home technology may allow lights, heating, doors, televisions and other devices to be operated using voice commands or assistive technology.

For some people with cognitive impairments, learning disabilities or neurodivergent conditions, AI may help simplify complicated information, organise tasks or explain instructions in different ways.

The key principle should be that technology supports independence rather than taking control away from the individual.

Accessible Education and Learning

Artificial intelligence could transform education for disabled students.

Traditional education systems often expect every student to learn, communicate and complete work in approximately the same way. AI can help provide more flexible approaches.

For example, students may be able to use AI tools to:

• Convert lecture recordings into searchable notes;
• Generate captions and transcripts;
• Explain difficult concepts using simpler language;
• Translate information into different formats;
• Create revision questions;
• Turn written material into audio;
• Organise study schedules;
• Provide spelling and grammar assistance;
• Support students with dyslexia or other learning differences.

AI should not replace teachers, lecturers or disability-support services, but it can provide an additional layer of personalised assistance.

Educational institutions must also ensure that students are not disadvantaged simply because they rely on assistive AI technology as a reasonable adjustment.

Employment and the Future of Work

Remote working, digital platforms and artificial intelligence have already changed employment opportunities.

For many disabled people, traditional workplaces can present barriers including inaccessible buildings, inflexible working hours, difficult commutes, sensory environments and employers who fail to understand reasonable adjustments.

AI-assisted work could enable more people to work remotely, become self-employed or establish businesses.

Possible opportunities include:

• Content writing and editing;
• Research;
• Digital marketing;
• Administration;
• Transcription;
• Data analysis;
• Graphic design;
• Website development;
• Customer support;
• Legal research assistance;
• Translation;
• Tutoring;
• Virtual assistance;
• Social media management;
• Business consultancy.

AI can assist with repetitive tasks, research, organisation, communication and productivity, potentially enabling people to concentrate on work that best suits their abilities.

However, accessibility must also apply to recruitment.

Employers should be cautious when using automated recruitment systems, CV screening software, online assessments or AI video-interview systems.

If these systems have been trained using narrow assumptions about communication, facial expressions, employment history or behaviour, disabled candidates may be unfairly disadvantaged.

Human oversight remains essential.

AI and Healthcare

Artificial intelligence is increasingly being explored within healthcare, including medical imaging, disease detection, monitoring systems, administrative support and personalised medicine.

For persons with disabilities, AI could help healthcare professionals identify problems earlier, analyse complex medical information or monitor changes in health.

AI-powered rehabilitation technologies, robotic devices and intelligent prosthetics may also improve mobility and independence.

However, healthcare AI must never reduce people to data points.

Doctors and healthcare professionals must continue to consider the individual patient’s circumstances, preferences and lived experience.

There must also be safeguards against discrimination if algorithms rely on historical datasets that contain inequalities or under-representation of disabled populations.

Communication Without Barriers

Communication technology is another area in which AI could have enormous benefits.

Modern AI systems can generate captions, convert speech to text and text to speech, translate languages and assist people who communicate differently.

For individuals who cannot communicate easily through speech, AI-assisted communication devices may help predict words or sentences based on context.

People with conditions affecting speech, mobility or dexterity may also benefit from eye-tracking, gesture recognition or other alternative methods of controlling technology.

The aim should always be to give people more ways to communicate, rather than forcing everyone to communicate according to one technological standard.

Accessibility Must Be Built In, Not Added Afterwards

A common mistake in technology development is treating accessibility as something that can be added after a product has already been created.

Inclusive technology works best when accessibility is considered during the earliest design stages.

Developers should consider questions such as:

• Can the system be used with a screen reader?
• Can it be operated without a mouse?
• Are captions available?
• Can users change text size and contrast?
• Is information presented in plain language?
• Can someone who cannot speak use the service?
• Does the technology work with assistive devices?
• Are users given alternatives if the AI system does not work for them?

Accessibility should not be considered a specialist feature used by a small minority.

Many accessibility features benefit everyone.

Voice typing, captions, predictive text and text-to-speech technologies are examples of features originally associated with accessibility that are now widely used across society.

The Importance of Involving Persons with Disabilities

Perhaps the most important principle in developing inclusive AI is straightforward:

Disabled people must be involved in designing it.

Technology created for disabled people without meaningful consultation can easily misunderstand what users actually need.

Persons with disabilities should therefore be represented in:

• AI development teams;
• User-testing programmes;
• Research projects;
• Accessibility advisory panels;
• Policy discussions;
• Government consultations;
• Technology companies;
• Academic research.

The principle often expressed within the disability-rights movement, “Nothing About Us Without Us”, is particularly relevant to artificial intelligence.

People who experience barriers every day are often the people best placed to identify where technology succeeds and where it fails.

The Risk of Algorithmic Discrimination

Artificial intelligence learns from data.

If that data reflects historical discrimination, exclusion or stereotypes, AI systems can potentially reproduce those patterns.

For example, automated systems could inadvertently disadvantage people because of:

• Disability-related gaps in employment;
• Unusual speech patterns;
• Facial differences;
• Mobility limitations;
• Communication differences;
• Medical history;
• Reliance on assistive technology.

Decisions concerning employment, education, healthcare, insurance, financial services or access to public services should therefore never be made blindly by algorithms.

Organisations should regularly assess AI systems for discriminatory outcomes and provide meaningful ways for individuals to challenge automated decisions.

Privacy and Sensitive Information

Many assistive technologies rely on highly personal information.

A person might use AI to process medical correspondence, describe their surroundings, communicate using assistive technology or manage disability-related appointments.

Organisations must therefore consider privacy and data protection carefully.

Users should understand:

• What information is collected;
• Why it is collected;
• How long it is stored;
• Whether it is shared;
• Whether it is used to train AI systems;
• How they can delete or correct their information.

Disabled people should not have to surrender excessive amounts of private information simply to access an accessible service.

Affordability and the Digital Divide

AI accessibility tools will only be genuinely transformative if people can afford to use them.

Some advanced assistive technologies can be expensive, while premium AI subscriptions, specialist equipment and high-speed internet access may be beyond the budgets of many disabled people.

Governments, technology companies, educational institutions and employers should consider how accessible technologies can be made available regardless of income.

Otherwise, society risks creating a new digital divide where people who can afford advanced AI assistance gain greater independence while those who cannot are left behind.

Human Oversight Must Remain

AI can support decision-making, but it should not automatically replace human judgement.

This is particularly important when decisions affect someone’s rights, welfare, employment, healthcare, benefits or access to essential services.

Individuals should be able to speak to a human being where necessary and should have clear ways to challenge decisions made or influenced by automated systems.

Accessibility means providing options, not forcing people into entirely automated processes.

Designing AI for Everyone

Building an inclusive future means recognising that disability is part of human diversity.

Technology should be flexible enough to accommodate different ways of seeing, hearing, communicating, moving, learning and interacting.

Developers, businesses and governments can contribute by:

• Adopting accessibility standards;
• Testing technology with disabled users;
• Carrying out equality-impact assessments;
• Improving AI transparency;
• Providing human alternatives;
• Ensuring compatibility with assistive technologies;
• Training staff in disability awareness;
• Monitoring systems for discriminatory outcomes;
• Offering accessible customer support;
• Publishing accessibility information clearly.

Accessibility should be viewed as part of good product design rather than an obligation that must reluctantly be met.

AI Should Empower, Not Exclude

The true measure of artificial intelligence will not simply be how fast it can process information or how many tasks it can automate.

Its success should also be judged by whether it improves people’s lives.

AI has extraordinary potential to increase independence, employment opportunities, education and participation for persons with disabilities. It could remove barriers that have existed for generations.

But that future is not guaranteed.

If accessibility, affordability, privacy and human rights are ignored, technology could reinforce inequality instead of reducing it.

The challenge for policymakers, developers, employers and society is therefore to ensure that innovation and inclusion develop together.

Conclusion

Artificial intelligence offers an opportunity to rethink accessibility.

Instead of expecting disabled people to continually adapt to systems that were not designed with them in mind, society now has the opportunity to design technology capable of adapting to people.

An accessible future means creating AI that recognises different abilities, communication methods and individual circumstances.

It means involving persons with disabilities in decision-making rather than designing systems on their behalf.

Most importantly, it means recognising that accessibility is not simply about disability.

When technology becomes easier to use, easier to understand and more flexible, everyone benefits.

The future of artificial intelligence should therefore not simply be intelligent.

It should be inclusive.

Important Note

AI should be viewed as an assistive tool rather than a replacement for human support, healthcare professionals, carers or qualified advisers. Accessibility, pricing, device compatibility and regional availability can differ, and AI systems can occasionally misunderstand information or produce incorrect answers. Disabled people should therefore retain control over how these technologies are used and should not be forced into AI-based systems as the only way of accessing essential services.

Further Reading & Resources

• https://unric.org/en/building-an-accessible-future-for-all-ai-and-the-inclusion-of-persons-with-disabilities/
• https://nothing-without-us.org/our-name-our-commitment/
• Seeing AI Microsoft: Designed specifically for blind and visually impaired people, Seeing AI uses artificial intelligence and a smartphone camera to describe the surrounding world. It can read printed text, describe photographs, recognise products and provide information about objects encountered in everyday life. It is free. https://www.seeingai.com/
• Be My Eyes AI: provides AI-generated descriptions of photographs and surroundings for people who are blind or have low vision. Users can photograph a menu, document, product, appliance or other object and ask follow-up questions about what is shown. Human volunteer assistance remains available when AI is not sufficient. https://www.bemyeyes.com/bme-ai/
• Google Lookout: uses a phone’s camera and sensors to help blind and visually impaired people identify objects, recognise text and understand their surroundings. Its reading tools can also make documents easier to access through adjustable text size, contrast, spacing and spoken output. https://support.google.com/accessibility/android/answer/9031274
• Google Live Transcribe: Particularly useful for Deaf and hard-of-hearing people, Live Transcribe converts spoken conversation and environmental sounds into text displayed on an Android device in real time. It can help with conversations, appointments, meetings, education and other situations where hearing speech may be difficult. https://support.google.com/accessibility/android/answer/9158064
• Apple Personal Voice & Live Speech: Personal Voice uses on-device machine learning and AI to create a synthesised voice resembling the user’s own voice. It can be particularly valuable for people who have, or are at risk of developing, conditions that affect speech. When combined with Live Speech, users can type what they want to say and have their device speak it aloud during conversations and calls. https://www.apple.com/uk/accessibility/
• Goblin Tools: Goblin Tools is a collection of simple AI tools developed largely with neurodivergent users in mind. Its Magic ToDo feature can break overwhelming jobs into smaller steps, while other tools can turn a brain dump into actions, estimate how long something may take, explain complicated information and help rewrite communications. https://goblin.tools/
• Speechify: Speechify converts written material into natural-sounding AI speech. It can read webpages, PDFs, documents and other text aloud and provides features such as text highlighting, adjustable reading speeds and voice typing. This can be useful for people with dyslexia, visual impairments, reading difficulties, fatigue or cognitive disabilities. https://speechify.com/
• Otter.ai: Otter uses AI to transcribe conversations and meetings into written text and can generate summaries, action points and searchable notes. It may be helpful for Deaf and hard-of-hearing users as well as people with memory, concentration, fatigue, executive-function or note-taking difficulties. https://otter.ai/
• ChatGPT: can support people with a wide variety of accessibility needs by simplifying complicated information, explaining documents in plain language, helping draft communications, breaking tasks into manageable steps, brainstorming solutions and using voice interaction where available. Voice allows users to speak and receive spoken responses while also following the conversation in text. AI-generated information should still be checked where decisions involve health, law, finances or personal safety. https://chatgpt.com/