Designing Accessible AI Curricula for Persons With Disabilities — evidence from a field pilot at Asha Kirana School for the Blind, Chikkamagaluru, Karnataka. Towards Atmanirbhar Bharat through Inclusive Digital Empowerment.
The gap between persons with disabilities and digital participation in India is not a gap in ability. It is a gap in design. Infrastructure built without them, curricula written around them, platforms that assume a standard set of senses and capabilities that millions do not share.
The NCPEDP 2022 survey found that 68% of employers cite skill mismatch as the barrier to employing persons with disabilities — not attitudinal resistance. The skills gap is real. And AI literacy is now one of its most consequential dimensions.
Free voice AI, screen readers, live transcription, and AI tutors make structured AI education feasible at near-zero cost. The window is open. The infrastructure question is whether we design it for everyone — or once again for only some.
"The gap is not inability — it is inaccessible infrastructure."
— Goudhaman & Srivastava (2026) · AI for Every Mind · CPP Conference, IIM BangaloreThe largest population of persons with disabilities in any single country — most without access to accessible digital learning infrastructure.
Census 2011 · MOSPI NSS 76th Round (2021)
A system-level failure. Not a failure of the students.
NCPEDP Survey (2022)
The 31-point gap is not explained by capability. It is explained by the absence of skilling infrastructure designed with PwD from the beginning.
PLFS 2020-21
Founded 1990 · Dr J.P. Krishna Gowda · Kempnalli, Malenadu Hills, Karnataka · 49 students, Classes 2–10
Blue = post-Gr 10 pilot cohort (n=9) · Grey = in-school curriculum (n=40, Cl 2–9)
Most accessible education is adapted — taken from a standard curriculum and modified. Both tracks here were built from the ground up, with the learner's sensory and cognitive reality as the starting condition, not an afterthought.
Framework: Universal Design for Learning (CAST 2018) · Social Model of Disability (Oliver 1990; Shakespeare 2006) · Sen Capability Approach (1999)
Voice-First · Screen Reader Compatible
Visual-First · Zero Audio Dependency
The insight behind this curriculum design is simple: children connect new things to what they already love. Every module meets students where they already are — in cricket, in music, in chess — and builds AI literacy from there.
Tools used: ChatGPT · Lichess · Wolfram Alpha · Suno AI · GitHub Copilot · Audacity+NVDA · eSpeak · IGNOU distance learning pathways
Modular, self-paced delivery with live Zoom-facilitated sessions — allowing students to work at their own pace while having direct access to the course designer for questions and walkthroughs.
The course designer delivered the programme directly — live demos, tool walkthroughs, and open Q&A sessions. Direct access to the designer ensured real-time adaptation to the class's needs.
School desktop computers and students' personal mobile phones were both used throughout. Curriculum was designed to work on both device types with equal accessibility.
On-campus training sessions for school staff were delivered by Dr. Goudhaman before student delivery began — building sustained in-school capacity beyond the pilot period.
Teacher observation notes and facilitator reflections during live Zoom sessions — qualitative, descriptive evidence. No psychometric instruments were used in Version 1 of the pilot.
Formal certificates were issued to all participants via the Divyang Art & Aaashraya Foundation, initiated by Programme Coordinator Shashikala — making completion tangible and official for every student.
This is descriptive evidence from teacher observation and facilitator notes — not a controlled evaluation. Its purpose is to document what the pilot made possible, and to ground the next phase of design in what actually happened.
Without being directed. Without prompting. Seven of the nine post-Grade 10 students navigated to ChatGPT and used it for their own purposes after completing only the second module — before the programme was half over.
Observed during live Zoom sessions · Facilitator notes, 2025
Five students engaged most deeply with the Music and Teaching modules. Two named a specific post-school goal for the first time in the programme — a career direction they had not articulated before.
Teacher observation notes · Asha Kirana School, 2025
Three students navigated to Lichess during an unstructured break — uninstructed, on their own initiative. The facilitator observed this as evidence of genuine absorption rather than curriculum compliance.
Facilitator reflection · Asha Kirana School, 2025
Every student completed the full curriculum on both desktop and mobile — without any technical support needed at any point after initial orientation. The design assumption was tested and held.
+ Teachers separately observed improved Science engagement: students began prompting AI for diagram descriptions independently.
These are not representative quotes selected for impact. They are the voices of the people who participated — a student, a principal, and a programme coordinator — each reflecting on what they witnessed.
"For the first time, I could understand a science diagram without waiting for someone to explain it to me."
"This programme had a huge impact on the confidence of our students. For the first time, they felt at par with the technological advancements happening around them — they were not behind, they were part of it."
"Seeing the students receive their certificates was deeply moving. Making it tangible and official for every student — watching them hold something that said they had mastered AI — made it real for all of us."
These are anticipated outcomes — not yet measured. They are grounded in observed behaviour during the pilot, curriculum design logic, and the Indian labour market context for persons with disabilities.
These are not gaps in aspiration — NEP 2020 and the RPwD Act 2016 both commit to inclusive education. They are gaps between commitment and implementation. Each one was a design constraint the curriculum had to solve.
No PMKVY Qualification Pack for persons with disabilities includes AI content. All existing modules predate the generative AI era (pre-2020) — making them structurally unable to prepare PwD for current labour market demand.
NSDC PMKVY 4.0 · NEP 2020 Para 6.11Government-funded platforms — Diksha, SWAYAM, PM eVidya — lack mandatory WCAG 2.1 compliance. They are inaccessible by design, not by accident. Inclusion was never a requirement in the platform specifications.
GIGW WCAG 2.0 · MeitY/DEPwD guidelinesSpecial educators lack AI tool training. No Diksha module exists for AI-inclusive classroom delivery. Teachers who want to use these tools have no institutional pathway to learn how — and those without access to the internet have even fewer options.
Diksha platform · NCERT teacher training frameworkPwD skilling programmes rarely connect graduates to employers specifically leveraging AI-intensive, remote, and text-based job roles — precisely the roles most accessible to persons with mobility or sensory disabilities. The skilling and the employment pipeline are disconnected.
RPwD Act 2016 · NCPEDP Survey 2022Each recommendation comes with a cost estimate and an implementation pathway. The argument here is not that this is a worthy cause. It is that the infrastructure already exists — it simply needs to be directed.
Create an 'AI Productivity Associate' qualification at NSQF Level 3–4. Target: 500,000 PwD students over 3 years. Estimated cost: ₹4,800/trainee via existing PMKVY infrastructure.
~₹240 Cr over 3 years · NSDC unit cost modellingMeitY/DEPwD directive requiring Diksha, SWAYAM, and PM eVidya to achieve WCAG 2.1 AA compliance within 24 months. Asha Kirana designated as the national accessibility test site.
MeitY / DEPwD directive · 24-month implementation timelineCertify 100,000 special educators per year in accessible AI tools: Seeing AI, Live Transcribe, AI tutors, and screen-reader compatible coding environments.
Diksha platform · NCERT · 100K educators/yearDPIIT/Startup India seed grants for accessible AI innovation: ISL-to-text AI, Braille-to-speech conversion, multilingual accessible tutors, and atypical speech recognition systems.
₹500 Cr · DPIIT / Startup India · AI Accessibility FundCommission 10 million annotated Indian Sign Language sequences under the National Language Technology Mission — enabling real-time ISL ↔ text AI. This directly supports the May 2026 Karnataka Deaf school pilot and all future scale. India has more than 2.7 million deaf citizens. A national ISL corpus is the infrastructure that makes accessible AI education possible at scale for this community.
10M annotated ISL sequences · NLTM · Directly supports May 2026 Deaf pilotStructured, accessible AI skilling produces measurable labour market outcomes when institutional infrastructure supports delivery. India must now build that infrastructure.
AI literacy transition programme measured over 3 years, 2020–2023 — demonstrating sustained, not temporary, employment impact.
Perkins School for the Blind (2023)
AI-for-employment programme, 2022–2023. Seven in ten participants in paid work within six months of programme completion.
RNIB (2023)
'Tech Pathways' captioned AI programme, Year 2 (2022) — 41-point improvement in digital competency, measured against baseline.
Deaf Society Australia (2022)
"Self-reliance is not the privilege of the able-bodied. It is the birthright of every mind."
Goudhaman & Srivastava (2026) · AI for Every Mind · CPP Conference, IIM Bangalore
Open to All
AI for Every Mind is freely available to any school, NGO, or government body.
Open to collaboration with institutions, NGOs, and policy bodies.
Goudhaman, S., & Srivastava, R. (2026). AI for Every Mind: Designing Accessible AI Curricula for Persons With Disabilities — Evidence from a Field Pilot at Asha Kirana School for the Blind, Chikkamagaluru. CPP Conference 2026. IIM Bangalore.