Navigating Challenges and Opportunities of AI in Supporting Disabilities in the Indonesian Workplace

  • 23/09/2024
  • Category: Commentaries
No Comments

Author: Tane Hadiyantono
Editor: Hafiz Noer

The exclusion and marginalisation of people with disability in the workforce often stem from physical limitations or communication barriers that impede them from interacting seamlessly. Yet, with the dawn of AI, these limitations could be bridged with generative AI to aid their task and communication. 

The main challenge will be to eliminate barriers in the work environment and public spaces that hamper people with disability from accessing AI. To solve this, three foundational steps that may help stakeholders to facilitate employment for persons with disabilities, are1: 1) Expand awareness; 2) Focus on the benefits; and 3) The adherence towards ethical principles. This article discusses foundational strategies that Indonesia could pursue to create an inclusive working environment for persons with disabilities. 

Facts and figures of disability at work in Indonesia

According to research by a policy advocacy group Prakarsa, Indonesia has 17.74 million people with disabilities who are in the productive age range with only 43% of them employed and 79% of them working at the informal sector2 (farming, fishing, working without contract at factories or retail sector). Over 60% are primary school dropouts and primary school graduates, and only 14% finished high school. 

Of those who work, 70% have a visual disability, 11% with hearing disability and more than half of the disabled workers have multiple disabilities.3 People with physical impairment that disables their hands, fingers, and communication are among the highest without a job while people with leg impairment could still work on wheelchairs.

Under the  Law  Number  8  of  2016  concerning  Persons with Disabilities4 place of work are mandated to ensure people with disabilities can utilize new technology to assist them in performing their duties; and for them to be able to freely choose their assistants by their own choices. There should be no limitations or discrimination against people with disabilities to obtaining employment administered by the government or private sector. The law also obliged government bodies and state-owned companies’ to employ at least 2% persons with disabilities and 1% for private companies. Thus, generative AI technology with its many features could become a great addition to the new technologies for people with disabilities at work. 

To push the point further, Indonesia has an open reception to using AI at work. In a survey conducted by Schwartz Reisman Institute for Technology and Society (SRI) and the Policy, Elections, and Representation Lab (PEARL) with more than 23,000 respondents from 21 countries, Indonesians ranked highly as citizens who are most optimistic about AI adoption, believing that it offers more benefits than drawbacks, particularly in speeding up tasks.5 The fact that the government launched the National Strategy on Artificial Intelligence (Stranas KA) in 2020 also highlighted the country’s interest in moving forward with AI. The Stranas KA will prioritize five national sectors comprised of health, bureaucracy reform, research and education, food security, mobility, and smart cities.6 

However, despite the high number of people who will benefit from the technology, there are limited generative AI apps for people with disabilities that are available in Indonesia. At the moment, there remains only one app by the Danish-origin ‘Be My Eyes’ that connects blind and low-vision individuals with sighted volunteers.7 There are several apps made by university students that won innovation competitions such as Katakan AI by Telkom University students8 that helps deaf people and hard of hearing communicate more seamlessly in virtual meeting rooms through voice-to-text, sign language interpreters, and automatic caption. Another example is AiCCESS by Universitas Brawijaya students9 created an app with voice-to-text and text summaries aimed at helping people with hard of hearing and deaf. However, both Katakan AI and AiCCESS never made it into public use, perhaps due to lack of access to funding. 

As shown above, there are AI technologies and innovations that could help people with disabilities in Indonesia. These could be incorporated to improve the work and productivity of people with disabilities. Indonesia could eliminate barriers in the work environment and the public space that hamper people with disability from accessing AI by adopting the following three foundational steps. 

Expand awareness: from disabilities to accessibilities 

As stated above, Indonesia has a large base of people with disabilities in the productive age with limited education and lack of formal employment. It is not only a missed opportunity but also a slight against the government’s welfare system, which is intended to treat its citizens with equal care. While regulations on equal employment10 and education11 are already in place, the reality is still far from ideal. However, to start identifying which group of persons with disabilities could benefit from the technology, we must first understand the most prevalent challenges. 

In this case, the World Bank identifies that public spaces and working environments where traditional written or verbal communications are often the only forms of communication available,12 became the biggest disadvantage to a person with visual, hearing, and neurodevelopmental disabilities. This situation hampers their ability to perceive and relay information seamlessly. 

This group requires assistive technology to aid their communication. Generative AI technology can help them communicate with speech-to-text, virtual environment, audio notetaker, word prediction software, spell checkers, gesture input, and other features that help face-to-face or remote interactions.13 

Through these technologies, a person with visual impairments can use Apple’s VoiceOver14 or Microsoft Seeing AI15 technologies to describe digital images or text. Meanwhile, those with speech impairment could use Google’s Parrotron16 to read their distorted speech pattern, and as mentioned before, there’s also the Be My Eyes app. 

Yet despite research and development efforts, the actual burden in actualizing AI technology in the workplace would be in the workplace’s commitment to making their company inclusive and accessible for persons with disabilities. Without putting inclusive design and participation of persons with disabilities into the workflow the human plus machine collaboration17 will bear risks of exclusion and biased information.

Focus on benefits: putting humans first

Problems like the cost of research and infrastructure overhaul to accommodate AI technology, and concerns about AI taking over jobs, often become the main points of negative discourse about AI in the workplace.

Yet by being focused on the negative, progress will remain stagnant. Instead, the state and decision-makers should concentrate on how AI technology could improve the hiring process for people with disabilities, enhance life at work, and optimise employee evaluations.18

There have been reports that companies are using AI hiring algorithms tools to filter applicants with disabilities.19 Such cases should be tackled with discriminatory charges and policy changes. On the other hand, AI technology needs to be trained to interact with persons with disabilities without bias or visual judgments, and to allow for more time flexibility.20 This approach would enable assessments to provide an analysis purely based on intellectual capacity and productivity probabilities. Could also navigate potential pitfalls by allowing future employees to map AI assistive technology that they need and how much it would contribute to their work. More information generated by AI and the use cases developed for persons with disabilities could also lead to the development of better machine learning models and large language models specifically designed for them. This could create a repository of valuable information resources, leading to better AI technology for disabilities.

Ethical principles to support persons with disabilities

AI ethical guidelines could help policymakers adopt an inclusive AI framework. In the case of AI for disability, responsible innovation21 could be a great building block. The framework highlights four ethical dimensions in technology, which comprises:1) Anticipation through contingency plans; 2) Reflexivity by owning the technology limitation; 3) Inclusion through involving people directly affected by the technology; and 4) Responsiveness of a company towards the latest changes. Similar to this framework is R(AI)se method22 which focuses on the responsivity, accessibility, inclusivity, and security of AI tools in the workplace. 

Companies should prepare contingency plans for the what-ifs scenarios when AI technology backfires. It could be in the form of biased algorithms, physical accidents with the gadgets, changing regulations, and all kinds of trouble. By preparing these plans, stakeholders could identify their shortcomings and work to solve the challenges. 

Including people with disabilities in the contingency draft plan as well as the operation workflow will provide direct insight on what works and what doesn’t work. And finally, by being prepared, companies will be able to navigate future challenges. Whereas at the government level, the OECD’s AI Principles is a good start for the government to design AI policies by ensuring an agile transition from the research and development stage to the deployment and operation stage for trustworthy AI use cases.23

  1. Accenture. (n.d). AI for disability inclusion. Accenture. https://www.accenture.com/content/dam/accenture/final/a-com-migration/custom/_acnmedia/pdf-155/Accenture-AI-For-Disablility-Inclusion.pdf.  (Accessed: 11 August 2024). ↩︎
  2. Perkumpulan PRAKARSA. (2022). Penyandang Disabilitas di Tempat Kerja: Kondisi dan Tantangannya di Indonesia sebagai Negara G20. Perkumpulan PRAKARSA. ↩︎
  3. Ibid. Prakarsa, p.25. ↩︎
  4. Republic of Indonesia. (2016) Law of the Republic of Indonesia Number 8 of 2016 on Persons with Disabilities., Database Peraturan Perundang Undangan. Available at: https://peraturan.go.id/files2/uu-no-8-tahun-2016_terjemah.pdf (Accessed: 11 August 2024). ↩︎
  5. Maslej, N. et al., (2024). The AI Index 2024 Annual Report. Institute for Human-Centered AI, Standford University. https://aiindex.stanford.edu/wp-content/uploads/2024/05/HAI_AI-Index-Report-2024.pdf (Accessed: 11 August 2024) ↩︎
  6. BPPT. (2020). Strategi Nasional Kecerdasan Artifisial Indonesia 2020-2045. Sekretariat Nasional Kecerdasan Artifisial Indonesia. https://ai-innovation.id/strategi (Accessed: 11 August 2024). ↩︎
  7. Be My Eyes. (n.d). Menjadi mata bagi penyandang tunanetra dan penglihatan terbatas. Be My Eyes. https://www.bemyeyes.com/language/indonesian (Accessed: 15 Agustus 2024). ↩︎
  8. Indonesia News Center. (2023). Telkom University Students’ ‘Katakan AI’ Idea Prevails at Microsoft Indonesia – AI for Accessibility Hackathon 2023. Microsoft. https://news.microsoft.com/id-id/2023/06/22/telkom-university-students-katakan-ai-idea-prevails-at-microsoft-indonesia-ai-for-accessibility-hackathon-2023/ (Accessed 15 August 2024). ↩︎
  9. Admin. (2023). Inovasi AiCCESS, Prototipe Bantu Tuna Rungu. Sustainable Development Goals Center Universitas Brawijaya. https://sdgs.ub.ac.id/inovasi-aiccess-prototipe-bantu-tuna-rungu/ (Accessed 15 August 2024). ↩︎
  10. Republic of Indonesia. (2003). Undang Undang Republik Indonesia Nomor 13 Tahun 2003 tentang Ketenagakerjaan. Kementerian Perindustrian. Available at: https://kemenperin.go.id/kompetensi/UU_13_2003.pdf (Accessed: 07 August 2024). ↩︎
  11. Ibid. Republic of Indonesia. p6. ↩︎
  12. Hata, A. Wang, H. Yuwono, J. Nomura, S. (2023). Assistive Technologies for Children with Disabilities in Inclusive and Special Schools in Indonesia. The World Bank. https://documents1.worldbank.org/curated/en/099543306052328820/pdf/IDU01f2788e204497047d60a3ea05db5ca5d1a6b.pdf (Accessed: 07 August 2024). ↩︎
  13. Raja, D. (2016) Bridging the Disability Divide through Digital Technologies, World Bank. Available at: https://thedocs.worldbank.org/en/doc/123481461249337484-0050022016/original/WDR16BPBridgingtheDisabilityDividethroughDigitalTechnologyRAJA.pdf (Accessed: 07 August 2024). ↩︎
  14. Apple. (n.d). Accessibility. Apple. https://www.apple.com/accessibility/ (Accessed: 11 August 2024). ↩︎
  15. Microsfot. (n.d). Seeing AI. https://www.microsoft.com/en-us/garage/wall-of-fame/seeing-ai/. (Accessed: 11 August 2024). ↩︎
  16. Biadsy, F. et al., (2019). Parrotron: An End-to-End Speech-to-Speech Conversion Model and its Applications to Hearing-Impaired Speech and Speech Separation. Interspeech. https://arxiv.org/abs/1904.04169. ↩︎
  17. Ibid. Accenture. P3. ↩︎
  18. Ibid. Accenture. P9. ↩︎
  19. Roth, S. Delbos, B. (2024). AI is causing massive hiring discrimnation based on disability. The Hill. https://thehill.com/opinion/technology/4576649-ai-is-causing-massive-hiring-discrimination-based-on-disability/ (Accessed on 11 August 2024). ↩︎
  20. Fisher, L. Bonaccio, S. Connelly, E. (2024). AI-based tools in selection: Considering the impact on applicants with disabilities. Organizational Dynamics. Volume 53, Issue 1, 
    https://doi.org/10.1016/j.orgdyn.2024.101036
    ↩︎
  21. Stilgoe, J., Owen, R., & Macnaghten, P. (2013). Developing a framework for responsible innovation. Research Policy, 42(9), 1568–1580. https://doi.org/10.1016/j.respol.2013.05.008 ↩︎
  22. Ibid. Accenture. p26-27. ↩︎
  23. OECD. (2024). Shaping an enabling policy environment for AI (Principle 2.3). OECD. https://oecd.ai/en/dashboards/ai-principles/P12 ↩︎