Responsible AI Fellowship

Overview

The Responsible AI fellowship is a capacity building program for social impact organisations. It equips them with the knowledge frameworks, tools, and capacities to develop equitable and safe AI interventions in critical social sectors.

The first program of its kind in India, the Responsible AI fellowship helps establish long-term capacities for equitable and ethical AI use in India.

Rationale

Across critical social sectors like healthcare, education and agriculture, several NGOs, tech-for-good organisations, and social impact ventures are developing new AI-based interventions. Increasing investments from government and philanthropic organisations are also propelling the development of AI for Social Good.

Social impact organisations can play a unique role in ensuring the inclusive and effective use of AI for Social Good. They work closely with populations in need and understand their concerns and contexts. Incorporating AI allows organisations to enhance the efficiency and scale of their programs while also providing personalised and responsive development interventions to their program recipients.

However, as they work with vulnerable populations in low-resource contexts, they also have to be extra cautious in using AI, ensuring that the appropriate guardrails are in place. For example, integrating an LLM-based AI chatbot could improve the efficiency and responsiveness of their intervention. However, if the bot has not been adequately tested, it could also give users incorrect answers and contribute to real-world harm. Organisations also collect a vast amount of sensitive personal data, which, if not handled properly, can undermine people’s right to privacy. Similarly, if the training data used to build the AI application is not adequately representative, integrating AI into existing programs could accentuate gender, caste or other social inequities.

Many organisations have not established the internal resources and capacities to ensure AI's safe and ethical use. For example, many do not have internal policies for data management or frameworks for monitoring and evaluation. In other cases, organisations are unsure of the type of AI model they need to build and the technical team they require.

While there are many frameworks and guidelines for Responsible AI, most are not contextualised to the specific needs of organisations or the social realities of India. They remain at the level of principles and do not provide adequate guidance for action. They also do not support organisations in understanding how trade-offs should be managed or the resources they need to sustain responsible and ethical practices.

Programme Features

• Bespoke program based on organisational needs and capacities. Design of curriculum and identification of mentors based on a needs assessment of chosen program participants.
• Expert lectures from leading thinkers and practitioners in the field of Responsible AI. Lectures cover themes such as data ethics and governance, privacy and security, explainable and accountable AI, monitoring and evaluation, and policy regulation.
• 1:1 mentoring sessions to help organisations resolve specific challenges they face. Mentors are selected based on domain expertise, technical knowledge, and experience building actionable frameworks for Responsible AI.
• Opportunities for peer learning from other organisations in the cohort through in-person and virtual community meet-ups and case study presentations.
• Supporting knowledge materials and tools to support organisations post the fellowship program. This includes actionable tools, case studies, and guidance on implementing responsible AI across the development and deployment lifecycle.

Preview: The Practice Playbook on Responsible AI