Towards Responsible AI: Ethical Challenges and Solutions for a Sustainable World
摘要
The concept of artificial intelligence (AI) is quickly becoming a key tool in promoting sustainable development, and its application can bolster climate adaptation and mitigation, clean-energy shifts, precision agriculture, environmental governance, and healthcare and environmental health monitoring. However, these advantages cannot be discussed without problematic ethical, environmental, and governance issues. Algorithms are biased, opaque, and collect data everywhere; all of which jeopardize equity, accountability, privacy, and civil liberties—and particularly marginalized groups. Simultaneously, compute- and data-intensive AI poses material cost burdens throughout its lifecycle: increasing electricity and water consumption, extraction upstream effects, a shortened hardware lifespan, and growing e-waste. Since the AI systems are integrated into the more complex socio-technical infrastructures, over-reliance and tight coupling may enhance systemic risks and spread errors at scale. The chapter presents an overview of emerging research and practice to analyze AI as a facilitator and pressure for sustainability. We discuss high-impact applications in energy, cities, agriculture, and health, ethical risks and sustainable future, and governance strategies in global and regional terms, from a participatory paradigm to participatory practices of responding to context. Based upon these, we propose a mission of implementation that includes sustainable sustainability through design and Green AI throughout the life cycle; requires transparency, auditability, and human control over risk and impact; assess risk and impact for assessment, use regulatory sandboxes to learn and be accountable; institutionalize multistakeholder, interdisciplinary collaboration with focus on the needs of the community and intergenerational equity, aimed at achieving cross-sector commitment to a sustainable life cycle; and the need to use mutually collaborative practice across disciplines. We argue at our heart that it is not merely possible but an obligation of social justice and strategic value to align AI innovation with the social justice and planetary boundaries of resilient, fair, and indeed sustainable futures of all.