Inspiration
"Remember to look up at the stars and not down at your feet..." — Stephen Hawking
Stephen Hawking demonstrated that disability should never limit a person's ability to learn, create, or contribute. His work inspired us to ask a different question: what if AI could adapt to the individual, rather than forcing the individual to adapt to technology?
Many existing accessibility tools improve access to digital content, but they often do not help users with cognitive, sensory, or motor disabilities learn effectively or complete complex tasks independently. We wanted to build an AI-powered system that understands a person's accessibility needs first, then personalizes the way it teaches and guides them.
This vision became The Hawking Initiative.
What it does
The Hawking Initiative is an AI-powered adaptive learning assistant designed to personalize education for people with disabilities.
The platform begins with an accessibility diagnostic questionnaire inspired by the International Classification of Functioning, Disability and Health (ICF) framework. Rather than asking users to disclose medical diagnoses, the assessment evaluates functional abilities across areas such as:
- Vision
- Hearing and communication
- Cognitive accessibility
- Executive functioning
- Motor accessibility
- Attention and fatigue
- Learning preferences
The assessment generates an Accessibility Profile, which is then used by an AI agent to personalize the learning experience.
The learning agent can:
- Generate structured learning plans
- Break complex concepts into smaller learning units
- Adapt explanations based on user needs
- Recommend appropriate learning formats (text, visuals, videos, or interactive exercises)
- Reduce cognitive load through step-by-step instruction
- Continuously personalize future lessons using user feedback
The goal is to improve accessibility, independence, and confidence for learners with diverse accessibility needs.
How we built it
The Hawking Initiative combines modern AI technologies with accessibility-first software engineering.
Our system consists of four major components:
- Diagnostic Engine that collects accessibility information through an adaptive questionnaire.
- Accessibility Profile Generator that converts assessment responses into personalized learning recommendations.
- Learning Agent powered by large language models that generates adaptive lesson plans, flashcards, summaries, quizzes, and explanations.
- Adaptive Learning Interface that dynamically adjusts the learning experience based on the user's accessibility profile.
The frontend is being developed using Next.js, TypeScript, Tailwind CSS, and shadcn/ui, while the backend is designed to support personalized user profiles and future learning analytics.
Challenges we ran into
One of the biggest challenges was deciding what information the AI actually needs to personalize learning.
Instead of asking users to identify specific disabilities, we redesigned the diagnostic around functional abilities, making it more inclusive while respecting user privacy.
Another challenge was balancing personalization with simplicity. We wanted enough information to meaningfully adapt lessons without overwhelming users with a lengthy assessment. Designing an accessibility-first user experience that remains intuitive for people with diverse needs required multiple iterations.
We also learned that accessibility extends far beyond screen readers and font sizes—it includes cognitive accessibility, executive functioning, attention management, and personalized instructional design.
Accomplishments that we're proud of
- Designed an accessibility-first AI learning framework rather than a generic AI tutor.
- Developed an ICF-inspired diagnostic questionnaire that focuses on functional accessibility rather than medical diagnoses.
- Created an adaptive Accessibility Profile capable of driving personalized learning experiences.
- Designed a scalable architecture that separates assessment, personalization, and learning into independent components.
- Built a vision that can evolve into a platform supporting education, workplace training, and independent living.
What we learned
Building accessibility technology taught us that personalization is just as important as accessibility.
People with disabilities do not all learn in the same way, and simply making content accessible is often not enough. Effective assistive technology should adapt to the individual's strengths, preferences, and challenges.
We also learned the importance of designing with established accessibility principles in mind, including the ICF framework and inclusive design practices, rather than assuming that one solution works for everyone.
Finally, we discovered that AI has significant potential not only to answer questions, but also to personalize how knowledge is presented, making learning more inclusive and effective.
What's next for The Hawking Initiative
Our long-term vision is to develop The Hawking Initiative into a comprehensive adaptive learning platform for people with disabilities.
Future work includes:
- Expanding the diagnostic assessment to better align with clinical accessibility standards and the ICF framework.
- Integrating multimodal learning experiences, including speech, images, videos, and interactive exercises.
- Implementing real-time adaptation based on learner progress and engagement.
- Collaborating with accessibility experts, educators, and clinicians to validate the diagnostic process.
- Conducting user studies to evaluate improvements in learning outcomes, accessibility, and learner independence.
- Supporting integration with existing Learning Management Systems (LMS) and educational platforms to broaden accessibility in schools and universities.
Built With
- google-adk
- python
- streamlit
- visual-studio
Log in or sign up for Devpost to join the conversation.