As the Global South increasingly adopts AI as a tool to address an array of challenges, we must ensure this does not come at the cost of algorithmic harms, ethical lapses, or widening inequality. Recognising that regulatory responses do not adequately address the gaps between principles and implementation, we at Aapti Institute set out to better understand what Responsible AI (RAI) means in practice. 

For startups and developers in a highly dynamic and resource-constrained environment, abstract RAI principles can be hard to navigate. Through our first phase of research, the goal was to explore how Indian AI practitioners can embed RAI principles into their workflows without stifling innovation. Over 3 months, we consolidated evidence through a mixed-method approach of extensive desk research and 17 semi-structured interviews. The interviews lent an interdisciplinary lens to our work, bringing together AI developers, startup founders, policy and legal experts, researchers, and journalists. Our findings brought forth two main outputs:

1. A Primer designed as a guide for Indian developers looking to integrate RAI practices into their work.
   
2. A White Paper consolidating insights, challenges, and actionable recommendations for policy and practice.

Key Learnings

Our research and community engagement revealed that widespread adoption of RAI principles in India is hindered by fragmented knowledge, a lack of accessible tools, and an absence of a shared community. While there is an initial understanding of RAI, there are capacity challenges, a lack of prioritisation and incentives, and minimal tangible support systems. For developers to implement RAI in India, we realised we need to build infrastructure – social, technical, and institutional, to support responsible innovation from the ground up.

This includes:

• Creating a shared repository of tools and resources
  

• Establishing a community of practice for peer learning and exchange
  
• Offering hands-on training and mentorship opportunities
  
• Promoting collaboration across sectors and disciplines
  
• Encouraging adoption of certifications and external standards
  
• Boosting public awareness and engagement

What will our second phase look like?

We have modelled our next phase of work to cater to the practical side of embedding RAI into system design. Over the next year, through a practice-based approach, we hope to build out a community-led ecosystem to help empower smaller players in India implement and adopt RAI practices. 

We plan to do this through workshops, a living archive, and a community of practice.

• Workshops: We will host sessions across India, inviting AI developers, leadership, and technical teams. These workshops will use interactive tools, including a custom-designed RAI card game simulating ethical dilemmas and decision making. We have already conducted a successful preliminary workshop with a range of stakeholders, where a card game to engage with the principles of RAI was piloted. We plan to hold more sessions over the year to foster dialogue, peer learning, and practical engagement with the core principles of fairness, transparency, safety, and privacy.
  
• Living Archive: We are building an evolving and self-sustaining digital repository to host case studies, open-source RAI tools, regulatory updates, and workshop outputs, creating a central knowledge hub by and for the AI practitioner community.
  
• Community of Practice: Through these efforts, we aim to build a sustainable network that reduces the entry barrier for smaller startups and encourages cross-learning, enabling them to adopt RAI practices strategically.
  

Over the next year, we will document our efforts, assess impact, and share findings through a final white paper, as a step towards more inclusive and ethical AI in the Global South. We hope to create a scalable, community-centric model for Responsible AI rooted in India’s AI landscape