Leveraging AI to Better Enable Accessibility

 

Credit: fizkes / Shutterstock.com, cve iv / Shutterstock.com, and Kate Roesch / EDUCAUSE © 2020

In the previous discussion, we examined six ethical questions that professionals must consider when developing Artificial Intelligence (AI)  solutions. In this post, we extend the topic of AI towards emerging technologies and trends. The impact that AI has on accessibility holds great promise; however, the COVID-19 pandemic illuminates the need for better technology to assist vulnerable populations that struggle with environmental shifts. For example, in the wake of the pandemic, many organizations, such as educational institutions, made emergency shifts to transition traditional learning environments to online learning. While this type of change affects all students, the impacts are especially difficult for students with disabilities. In a recent EDUCAUSE Exchange podcast, Brewer et al. (2020) experts discuss AI for improving accessibility for higher education students with disabilities, including the challenges and emerging opportunities for AI.  The EDUCAUSE research community is the custodian of the annual Horizon Report, a forum for advancing higher education and technology choices (Brown et al., 2020). 

In a 2016 survey of undergraduate students, 19% of the respondents indicated having some type of disability, whether physically visible or invisible to others. Further, given the population had the choice not to answer disability-related questions, researchers believe the actual percentage is even higher (Gierdowski & Galanek, 2020). The key findings from the study as a result of the survey provide multiple opportunities. Still, one particular relates directly to AI: the desire for a more engaging classroom experience through interactive technology. In Brewer et al. (2020), the panelists elaborate on three specific areas where AI can better student independence and learning. 

First is improving the user experience, especially in Machine Language (ML) based speech synthesis. While the quality of synthetic speech is refining, the venture aims at becoming a more natural and conversational dialog. The second area involves the improvement of self-contained processes. AI technology must produce accessible content rapidly and in real-time as opposed to reliance on manual conversion techniques. For example, the future vision includes AI automatically describing images, something that is not near operational. Lastly, AI and ML's advancement provides new interaction methods, such as spoken dialog models for the visually impaired. The future of natural language interaction depends on advances in AI/ML technology, where research and commercial entities must work together to develop and produce technologies (Prajwal et al., 2019). 

Two forces that challenge the advancement are technical and socioeconomic, where the consumer marketplace is a driving factor. Large organizations such as Amazon, Microsoft, and Google all have products on the market that enable Natural Language Processing (NLP) through AI/ML-based technology. For example, the Amazon Alexa product is available on a wide range of devices and growing in popularity as a consumer offering. Amazon is currently in trial with new features that enable bi-model dialog between humans and computers, which is no trivial task (Metz, 2020). To reach the goals described in Brewer et al. (2020), advancements in the integration of AI, ML, and NLP is the critical component towards natural language interaction. The emergence in integrating the three components driven from multiple domains holds promise for improving technology in education for learners with a disability (Kang et al., 2020). 

References

Brewer, J., Gerard, C., & Hakkinen, M. (2020). EDUCAUSE Exchange In The Impact of AI on Accessibility. https://er.educause.edu/podcasts/educause-exchange/the-impact-of-ai-on-accessibility

 

Brown, M., McCormack, M., Reeves, J., Brook, D. C., Grajek, S., Alexander, B., Bali, M., Bulger, S., Dark, S., Engelbert, N., Gannon, K., Gauthier, A., Gibson, D., Gibson, R., Lundin, B., Veletsianos, G., & Weber, N. (2020). 2020 Educause Horizon Report Teaching and Learning Edition. https://library.educause.edu/resources/2020/3/2020-educause-horizon-report-teaching-and-learning-edition

 

Gierdowski, D. C., & Galanek, J. D. (2020). ECAR Study of the Technology Needs of Students with Disabilities, 2020. EDUCAUSE Review. https://www.educause.edu/ecar/research-publications/ecar-study-of-the-technology-needs-of-students-with-disabilities/2020/introduction-and-key-findings

 

Kang, Y., Cai, Z., Tan, C.-W., Huang, Q., & Liu, H. (2020, 2020/04/02). Natural language processing (NLP) in management research: A literature review. Journal of Management Analytics, 7(2), 139-172. https://doi.org/10.1080/23270012.2020.1756939

 

Metz, R. (2020). Why Amazon's new Alexa conversation feature is so hard to pull off. CNN Business. https://www.cnn.com/2020/09/25/tech/amazon-alexa-conversational-ai/index.html

 

Prajwal, S. V., Mamatha, G., Ravi, P., Manoj, D., & Joisa, S. K. (2019). Universal Semantic Web Assistant based on Sequence to Sequence Model and Natural Language Understanding. 2019 9th International Conference on Advances in Computing and Communication (ICACC), 110-115. https://doi.org/10.1109/ICACC48162.2019.8986173