We are the Non-Profit Working Group on AI (Singapore), a group comprising members of DataKind SG, Effective Altruism SG, AI researchers, data scientists, academics and more.
We are collectively interested in the impact of that AI will have on society in Singapore and beyond, and have organized to foster greater engagement with AI issues in the non-profit sector and in civil society.
Background: the PDPC’s Model AI Governance Framework
AI holds great promise to advance technology, increase efficiency and introduce revolutionary tools into our lives. However, AI and other data analytics systems have often been implemented without proper safeguards, and have exacerbated certain social ills. In recognition of the potential risks, the Personal Data Protection Commission (PDPC), a government agency, published “A Proposed Model AI Governance Framework” in January 2019. The Model Framework introduced some safeguards for the development and deployment of AI, and encouraged companies to adopt those safeguards in their internal governance. The PDPC called on the public to provide feedback, so as to improve upon the current proposed framework.
Response to the Model Framework
Answering the call for feedback, we have prepared a detailed paragraph-by-paragraph response to the model framework. In the interest of public transparency and accountability, we are making the response available here: Response to the Model AI Framework.
We hope that our response and the contributions in it can spark a conversation about AI principles and practices between civil society, academics, researchers, industry, and government.
Major themes from our recommendations are listed as follows:
- Clarifying and broadening the scope of AI governance
- Improving the explanation of harms and risks
- Improving the explanation of guiding principles
- Safeguards for both software practice and human decisions
- The social responsibility of companies
- AI ethics requires consultation with diverse voices
- The need for critical reflection beyond procedures and frameworks
Areas for Further Research and Governance
In addition, we have suggested the following additional areas for research and governance:
- The competitive dynamics between AI companies
- AI research undertaken by organizations
- The distribution of AI development, deployment, and monitoring across several companies