Response to the U.S. Department of Commerce Request to Information on AI and Open Government Data Assets

Open data is a key tool for government transparency and accountability. Thus, the improvement of processes to create, curate, distribute and maintain governmental open assets is crucial to safeguarding democracy. Over the past years, we saw the rapid dissemination of artificial intelligence (AI) systems worldwide, especially those based on large language models (LLMs) and other forms of generative AI that use deep learning models. The importance of open data assets to train such models has led different organizations to seek ways to make their data available to stakeholders pursuing innovative solutions for long-lasting, complex societal problems. However, besides the potential for technological innovation, there are several concerns, harms and ethical issues that arise with making data available and using them to train AI models. We must move beyond the idea that simply making various datasets available will ensure transparency; instead, we should seek to understand the needs and demands of individuals and organizations, to truly use open data assets for the public interest. Having clear goals, understanding the environmental impact context, addressing social implications of opening data, investing in a robust infrastructure and in data governance to make data available is a starting point of a long journey. The reason for such a complex process is because we think about open data as being technical instead of being about people and their needs. When planning to prepare open government data assets for the use of AI, the human and environmental aspects of this process should not be an afterthought. The world saw quick changes brought by the popularization of AI – systems that need massive volumes of data and resources – while open data around the globe have not advanced at the same speed. The U.S. scores relatively low in terms of data availability and use and impact, showing that as AI becomes the center of attention in this process, there are structural issues in open government data that need to be addressed. This document will focus on issues and considerations in (1) data ethics and digital rights, (2) data dissemination standards, (3) data integrity and quality, (4) partnerships and (5) climate action.

Reia, J. and Leach, R. (July 16, 2024) Response to the U.S. Department of Commerce Request to Information on AI and Open Government Data Assets (89 FR 27411).