Expanded ASEAN Guide on AI Governance and Ethics – Generative AI

Share on Social Media

The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI is a pivotal document that builds upon the foundational principles outlined in the ASEAN Guide on AI Governance and Ethics (2024). This expanded guide specifically addresses the opportunities, risks, and policy considerations related to Generative AI (Gen AI), a transformative technology that has rapidly evolved in recent years. The document aims to provide ASEAN policymakers with a comprehensive framework to promote the responsible adoption of Gen AI while mitigating its associated risks. Below, we delve into the key aspects of the guide, its policy recommendations, and the value it adds to the ASEAN region.

1. Context and Objectives

1.1 The Rise of Generative AI

Generative AI represents a significant leap in artificial intelligence capabilities. Unlike traditional AI, which focuses on classification and prediction, Gen AI is designed to create new content, including text, images, audio, video, and even computer code. This technology has the potential to revolutionize industries, enhance public services, and democratize access to AI tools for individuals and small businesses. However, its ability to generate realistic content also introduces new risks, such as the spread of disinformation, deepfakes, and intellectual property infringement.

The ASEAN region, with its diverse linguistic and cultural landscape, stands to benefit immensely from Gen AI. The guide estimates that the total economic opportunity for Gen AI across the Asia-Pacific region could reach nearly $6 trillion by 2038. However, to fully harness this potential, ASEAN must address the ethical, legal, and societal challenges posed by Gen AI.

1.2 Objectives of the Expanded Guide

The primary objective of this expanded guide is to supplement the ASEAN Guide on AI Governance and Ethics (2024) by providing targeted policy recommendations for the responsible adoption of Gen AI. The guide aims to:

Highlight the opportunities and risks associated with Gen AI.
Offer a set of policy actions to promote safe and ethical Gen AI use.
Encourage regional interoperability and harmonization of AI governance frameworks.
Foster innovation while ensuring that Gen AI systems are aligned with human values and societal well-being.

2. Key Risks of Generative AI

The guide identifies six primary risks associated with Gen AI, which were initially outlined in the ASEAN AI Guide (2024):

Mistakes and Anthropomorphism: Gen AI systems can produce highly coherent but incorrect outputs, often referred to as “hallucinations,” which can lead to harmful consequences, such as incorrect medical advice or flawed software code.
Factually Inaccurate Responses and Disinformation: Gen AI can amplify false or misleading information, eroding trust in reliable sources and shaping public perception in harmful ways.
Deepfakes, Impersonation, and Fraudulent Activities: The ability of Gen AI to create realistic content, such as deepfakes, poses significant risks for identity theft, fraud, and the spread of misinformation.
Infringement of Intellectual Property Rights: Gen AI systems trained on copyrighted works without proper authorization can lead to legal challenges, especially if the generated content closely resembles existing works.
Privacy and Confidentiality: Gen AI systems can inadvertently memorize and reproduce sensitive training data, leading to privacy breaches. Malicious actors can also exploit these systems to extract confidential information.
Propagation of Embedded Biases: Gen AI systems can inherit biases from their training data, leading to outputs that reinforce stereotypes or discriminate against certain groups.

In addition to these immediate risks, the guide also explores frontier and systemic risks, such as the potential for Gen AI to be used in creating biological weapons, the challenges of controlling highly advanced AI systems, and the environmental impact of increased computing power.

3. Policy Recommendations

The guide provides a comprehensive set of policy recommendations organized around nine key dimensions:

Accountability: Establishing shared responsibility frameworks across the Gen AI value chain, including developers, deployers, and users.
Data: Promoting data sharing and collaboration to develop ASEAN-relevant Gen AI models while ensuring compliance with data protection laws.
Trusted Development and Deployment: Encouraging best practices for the safe and ethical development of Gen AI systems, including transparency and user empowerment tools.
Incident Reporting: Developing frameworks for reporting and managing incidents related to Gen AI failures or misuse.
Testing and Assurance: Establishing regionally applicable benchmarks and testing tools to ensure the safety and reliability of Gen AI systems.
Security: Addressing unique cybersecurity challenges posed by Gen AI, including vulnerability detection and secure-by-design principles.
Content Provenance: Implementing technical solutions, such as digital watermarking, to distinguish AI-generated content from human-created content.
Safety and Alignment Research: Promoting research into the long-term safety and alignment of Gen AI systems with human values.
AI for Public Good: Encouraging the use of Gen AI to improve public services, enhance economic development, and address regional challenges.

4. Use Cases

The guide illustrates its policy recommendations with four detailed use cases from ASEAN countries, showcasing how public and private institutions are implementing AI governance and ethics in practice:

PhoGPT (Vietnam): An open-source large language model (LLM) developed by VinAI, tailored for the Vietnamese language and culture. PhoGPT aims to democratize access to Gen AI technology in Vietnam and other ASEAN countries.
Project Moonshot (Singapore): An open-source toolkit developed by the AI Verify Foundation to streamline the testing of LLM applications, ensuring their safety and cultural relevance.
Responsible AI Internal Programme (Accenture): A comprehensive governance framework implemented by Accenture to ensure the ethical use of AI across its operations in ASEAN.
ThaiLLM (Thailand): A public sector initiative to develop a Thai-language LLM, providing a shared national infrastructure for AI development and fostering innovation in Thailand.

These use cases demonstrate the practical application of the guide’s principles and highlight the potential for Gen AI to drive economic growth and social progress in the ASEAN region.

5. Value Added by the Expanded Guide

The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI adds significant value to the region in several ways:

Regional Coordination: The guide promotes a unified approach to AI governance across ASEAN, ensuring that member states can collaborate effectively to address the challenges and opportunities of Gen AI.
Pro-Innovation Framework: By emphasizing the importance of innovation and economic growth, the guide encourages the development of Gen AI technologies that can drive regional competitiveness.
Risk Mitigation: The guide provides a clear roadmap for mitigating the risks associated with Gen AI, ensuring that the technology is used responsibly and ethically.
Cultural Relevance: The guide recognizes the unique linguistic and cultural diversity of the ASEAN region, advocating for the development of Gen AI models that are tailored to local contexts.
Global Alignment: The guide aligns with global best practices and frameworks, such as those developed by the OECD, ISO, and NIST, ensuring that ASEAN remains interoperable with international standards.
Public Sector Empowerment: By highlighting the potential of Gen AI to improve public services, the guide encourages governments to leverage AI for the public good, enhancing citizen well-being and administrative efficiency.

6. Conclusion

The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI is a timely and essential resource for policymakers, industry leaders, and stakeholders in the ASEAN region. It provides a balanced approach to harnessing the transformative potential of Gen AI while addressing its ethical and societal challenges. By fostering regional collaboration, promoting innovation, and ensuring the responsible use of AI, the guide positions ASEAN as a leader in the global AI landscape. As Gen AI continues to evolve, this guide will serve as a foundational document for shaping the future of AI governance in the region and beyond.

This expanded guide not only reinforces ASEAN’s commitment to ethical AI development but also sets a precedent for other regions to follow in navigating the complex landscape of Generative AI.