Open Source AI: Policy Options for India

The past year has seen tremendous developments in open-source AI. While some AI companies have focussed on closed-source and proprietary approaches, others have open-sourced their AI models, data sets and codes. Open-source AI is believed to create a level playing field for AI development and achieve the Sustainable Development Goals. However, many debates and dilemmas remain unanswered - from the appropriate definition of open source in AI to questions of liability and accountability.

As policymakers and users worldwide grapple with rapid AI development, the economic benefits of AI are concentrated among those who have access to data, computing, and resources. Open-source AI provides a potential pathway out of this conundrum, but before that, many of these debates and dilemmas need to be addressed.

Moreover, it is essential to understand the positions of various stakeholders in the ecosystem and provide practical recommendations on the way forward. For example, this includes broader questions about the political economy of the ecosystem and more specific considerations on issues such as data hosting and selection of licenses.

To this end, this project aims to:

• Articulate the opportunities, challenges and tensions related to open-source AI
• Identify the key points of consensus and disagreement among critical stakeholders.
• Formulate actionable recommendations for key stakeholders across government, industry and civil society