Today marks the Global Accessibility Awareness Day (GAAD). GAAD is an awareness day focusing on digital access and inclusion for the more than one billion people with disabilities in the world. Given the growing importance of emerging technologies in our societies and economics, we dedicate this article on GAAD to accessibility of new technologies, particularly of Artificial Intelligence (AI) systems.
New technologies – new opportunities for persons with disabilities
New technologies, such as Artificial Intelligence (AI) can increase accessibility and independence for persons with disabilities. For example, incorporating AI in user interfaces and user interaction will be particularly useful for persons with sensory and cognitive disabilities, as the devices and operating systems can automatically adjust to the behaviour and needs of the user.
Other examples of AI supporting inclusion and accessibility of persons with disabilities, include AI which can increase accessibility of images by automatic alternative text, ‘seeing’ AI apps, which provide image description to those who are blind or partially sighted, AI used for automatic captioning, translation, AI-based sign language avatars, AI used to clarify and create visuals of complex written text, and others described in the EDF ‘Plug & Pray’ report on emerging technologies. A word of caution here is important, as many of these technologies are still in their early stages of development and need further perfection in order to allow persons with disabilities to realise their full potential as independent members of their societies.
New technologies – old barriers and new risks for persons with disabilities
However, rapid digitalisation and use of AI is not without risks for persons with disabilities. One of the risks is of course that inaccessible AI systems and their deployment can lock persons with disabilities out of social participation. Digitalisation in general, if not done with accessibility and human diversity in mind, can reinforce existing barriers, and create new ones. Additionally, there are risks associated with new technologies infringing on the human rights of persons with disabilities and further increasing inequality gaps. These technologies can also create safety risks for persons with disabilities if not developed with a design for all approach or don’t have accessible interfaces to terminate the operation of the technology in case of malfunction. Examples of inaccessible and disability-excluding AI-based tools are unfortunately not uncommon: for example a speech recognition system not able to understand commands made by a person with down syndrome, algorithm-based smart assistants or machine learning models for autonomous cars not recognising persons with disabilities as human, crowdsourcing websites for remote work inaccessible for persons with disabilities.
What the EU must do
At the moment, the EU is at a very opportune moment to ensure that AI systems are accessible for persons with disabilities, support their social participation, and do not risk causing harm to their fundamental rights. This is because, the EU is in the process of developing the EU AI Act to regulate the development and deployment of AI-based technologies.
EDF has been campaigning for the EU AI Act to include mandatory accessibility requirements for providers and users of AI systems, as well as measures to protect persons with disabilities from risks of harm to fundamental rights. Only by strong legal requirements on accessibility and safeguards against harm, can the EU ensure that persons with disabilities benefit equally from emerging technologies, instead of being left behind or harmed by use of AI systems.
Mher Hakobyan, EDF Accessibility Officer (firstname.lastname@example.org)
Alejandro Moledo, EDF Head of Policy (email@example.com).