How Microsoft’s AI For Accessibility Is Addressing The Issue Of Data Desert

AI-based technology has been very generalised and hence very exclusionary in catering to persons with disabilities.

The lack of machine learning datasets that include people with disabilities has proved to be a major hindrance in developing solutions customised to their needs. This phenomenon is often referred to as ‘data desert’. It is common practice for organisations to build technology products and services to use data at an aggregate level, leading to stereotyping and exclusion in the process. 

For the longest time, AI-based technology has been very generalised and hence very exclusionary in catering to persons with disabilities. It has led to the ousting of a considerable chunk of society from the benefits of AI. But now, Microsoft is leading the way with joint efforts by advocacy organisations that hope to do something for the inclusiveness of AI in this “data desert”.

ttps://twitter.com/MSFTEnable/status/1417159949466181636?s=20

Microsoft’s AI for accessibility

Microsoft announced the AI for Accessibility initiative in May 2018. The company had then pledged $25 million for AI development projects by universities, philanthropic organisations, and others to develop technology for people with disabilities.

Here are some of the recent AI for accessibility projects led by Microsoft to make AI more inclusive:

  • Braille AI Tutor: Microsoft Corp. has joined hands with Massachusetts-based startup ObjectiveEd in its efforts to improve educational outcomes for children with disabilities. ObjectiveEd’s Braille AI Tutor is a teaching aid for students with impaired vision. It uses gamification to help students practise independently or during distance learning via speech recognition with a braille display.
  • I-Assistant: inABLE, a Kenyan non-profit organisation, that empowers blind and low vision persons through computer assistive technology, has joined hands with India-based non-profit organisation I-STEM (it develops STEM technical solutions) to produce I-Assistant. It uses a custom-designed model, based on top of Azure Computer Vision to detect different structures and layouts for documentation and processed text, including math expressions. It can also create a readable and accessible document that converts any examination into a digitally accessible format. Users can navigate the test or paper smartly and perform background actions with Azure Custom Speech Recognition, Text to Speech, and Language Understanding (LUIS), such as checking the exam status. 
  • ADMINS chatbot: UK’s largest academic institution, Open University students have worked with professors to create an ADMINS chat assistant ‘Taylor’. This model makes forms programmatically accessible and also explores an alternative form completion process that closely matches a natural dialogue between a disabled student and a smart chatbot. The ADMINS team analysed forms and dialogue between students and expert advisers through the Azure Bot framework and speech services to recognise relevant intentions and to develop a system that promotes student engagement during a conversation, either spoken or typed. The chatbot uses the free form of the student’s responses with Azure Language Understanding (LUIS) to automatically lead the student through each step of the process and interpret answers to the corresponding forms.
  • TigerChat: Rochester Technology Institute (RIT) and National Technical Institute for the Deaf (NTID) have developed the TigerChat app to help improve communication for students who are deaf or hard-of-hearing.TigerChat was established with the support of CloudCheckr, a Rochester-based technology company, through Microsoft’s AI for Accessibility grant. Furthermore, RIT is exploring different ways of improvements in captioning options for Deaf and hard-to-hearing students, including automatic removal of disfluencies and addition of punctuation.

Roadmap for Data Oasis

While Microsoft is working towards AI inclusiveness and eradication of data desert, the absence of machine-learning datasets representing or including disabled individuals is a common barrier for researchers or developers in many sectors and enterprises. Especially those that are working to develop smart solutions to support daily work or AI systems for such groups. To ensure that all individuals have access to the benefits of data-driven innovation and no one is at a disadvantage, policymakers should also strive to eradicate data poverty and close the data divide from the ground up. This can also be achieved if companies start investing in AI-driven projects that reduce bias by including data on disabled people and make software and devices smarter and more contextually relevant, and cost-effective.

More Great AIM Stories

Ritika Sagar
Ritika Sagar is currently pursuing PDG in Journalism from St. Xavier's, Mumbai. She is a journalist in the making who spends her time playing video games and analyzing the developments in the tech world.

More Stories

OUR UPCOMING EVENTS

8th April | In-person Conference | Hotel Radisson Blue, Bangalore

Organized by Analytics India Magazine

View Event >>

30th Apr | Virtual conference

Organized by Analytics India Magazine

View Event >>

MORE FROM AIM
Yugesh Verma
All you need to know about Graph Embeddings

Embeddings can be the subgroups of a group, similarly, in graph theory embedding of a graph can be considered as a representation of a graph on a surface, where points of that surface are made up of vertices and arcs are made up of edges

Yugesh Verma
A beginner’s guide to Spatio-Temporal graph neural networks

Spatio-temporal graphs are made of static structures and time-varying features, and such information in a graph requires a neural network that can deal with time-varying features of the graph. Neural networks which are developed to deal with time-varying features of the graph can be considered as Spatio-temporal graph neural networks. 

Yugesh Verma
A guide to explainable named entity recognition

Named entity recognition (NER) is difficult to understand how the process of NER worked in the background or how the process is behaving with the data, it needs more explainability. we can make it more explainable.

3 Ways to Join our Community

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Telegram Channel

Discover special offers, top stories, upcoming events, and more.

Subscribe to our newsletter

Get the latest updates from AIM