Takshashila Discussion Document - A Pathway to AI Governance
This image was generated using an AI tool (DALL-E)
Published November 25, 2023
Executive Summary
Artificial intelligence has immense potential to enhance human capabilities and drive growth in several industries. It can also greatly improve education, healthcare, and governance outcomes, particularly benefiting low-income countries. However, this potential may not be realised if the AI market remains concentrated in the hands of a few dominant players.
While global AI governance efforts primarily focus on ethics and safety, it is crucial to consider AI governance from a national interest perspective. This involves examining how AI adoption can help humans flourish, strengthen democracy, and promote a stable global order. It also looks at the need for sustainable practices in AI adoption and the importance of ensuring competition in the AI ecosystem. These considerations need to be examined across these different stages of the AI supply chain - data, computation, model, and application - to envision the desired outcomes at each stage.
At the data stage, a marketplace that recognises individual ownership of data and empowers individuals to dictate the usage and distribution of their personal data is essential. Additionally, having datasets that accurately represent the target population for various use cases is vital to reduce algorithmic bias.
AI development relies on computing infrastructure, especially in the cloud, and promoting competition, reducing entry barriers, and preventing monopolistic practices in the AI cloud service provider market is essential. Developing domestic capacity in cloud infrastructure and AI chips is also important for strategic autonomy.
At the model stage, a competitive marketplace that fosters innovation and accessibility is essential, offering various distribution models from proprietary to open-source. Governments should also support open-source AI technologies with broad research and commercial applications.
Lastly, addressing concerns in other stages of the AI ecosystem will pave the way for the market to address the requirements of the application layer effectively. However, it is essential to adopt a risk-based framework to help strike a balance between proceeding cautiously in high-stakes AI applications and encouraging innovation in other areas.
Authors
