BOSTON UNIVERSITY SCHOOL OF LAW

Artificial Intelligence (AI) systems—encompassing generative AI, predictive AI, automated decision systems, agentic AI and content-moderation tools—are being deployed across many domains of public and private life. AI systems are celebrated for saving time and labor but exact high costs on the very institutions essential to the creation of stable, just, and prosperous societies necessary for attaining the Sustainable Development Goals envisioned in Agenda 2030. Evidence is growing that the rapid and indiscriminate deployment of AI systems could degrade the institutions democratic governance depends on to function in a free society. Institutions serve to structure complex human social interactions. Professors Hartzog and Silbey have argued that AI systems can erode institutions by undermining expertise, short-circuiting decision-making, and isolating humans. A successful dialogue would take on mitigating this risk.

Social, economic, ethical, cultural, linguistic and technical implications of AI

The Global Digital Compact recognizes that access to relevant, reliable, and accurate information and knowledge is essential for an inclusive, open, safe and secure digital space, and that digital and emerging technologies can enable the malicious manipulation of information in ways that are harmful to society. Professors Hartzog and Silbey have argued that AI systems risk undermining the important institutional role journalists play in maintaining the public sphere of information as "AI slop" pollutes the information ecosystem and devalues the expertise and legitimacy of journalism. Journalists play an institutional role in informing public debate, educating the public, and holding power accountable. Democracy requires informed deliberation and local civic life needs continuous human interaction. The more deliberation is delegated to AI systems and the more isolated we become from one another, the greater the risk to society.

The environmental impacts of AI infrastructure represent a rapidly growing human rights concern that demands urgent attention. AI's need for computing power to train data requires energy consumption and carbon emissions on a massive scale could ravage the environment. Major technology companies are investing hundreds of billions of dollars in AI data centers. Research from the Massachusetts Institute of Technology and the Harvard Business Review has documented the significant environmental impact of generative AI and the uneven distribution of those impacts. The Global Digital Compact explicitly commits states to promote sustainability across the life cycle of digital technologies, including context-specific measures to increase resource efficiency, to conserve and sustainably use natural resources, and to ensure that digital infrastructure is sustainably designed to address environmental challenges. The negative environmental impacts associated with AI systems are not evenly distributed with burdens falling disproportionately on lower-income populations, vulnerable populations, and communities of color at every stage of the life cycle of AI technology. The manufacture of AI computing hardware relies on the extraction of critical minerals often using mining practices that cause ecological damage. Improper disposal of AI computing hardware creates hazardous waste. AI data centers consume tremendous amounts of energy from fossil fuel sources contributing to climate change. Adverse environmental impacts associated with AI systems must be viewed through a human rights lens. The right to a clean, healthy, and sustainable environment is directly implicated by the energy-intensive infrastructure required for training and running large-scale AI models.

Government failure to enforce existing privacy, intellectual property, antidiscrimination and environmental laws that directly or indirectly regulate business enterprises can present risks to human rights. States should enforce and regularly assess the adequacy of laws and regulatory frameworks. In the United States communities hosting data centers often bear disproportionate environmental burdens, including increased water usage and local air quality impacts, while the benefits of AI systems flow predominantly to technology companies and their shareholders. This pattern reproduces the structural inequalities that both the UNGPs and the Global Digital Compact seek to address. To that end, we hope the dialogue will take seriously the environmental human rights issues that arise in the AI supply chain.

The AI Dialogue should connect with environmental justice organizations, technology and human rights organizations, with a focus on underrepresented and vulnerable populations.