Prioritizing Policies for Furthering Responsible Artificial Intelligence in the United States

Pre-deployment audits and assessments should be the top priority for government agencies and companies, as the paper identifies these as having the highest potential impact for ensuring AI systems work responsibly before they're released to the public, despite acknowledging significant barriers to implementation.

Post-deployment accountability mechanisms are critical and should be prioritized by the U.S. legislature, enabling oversight and corrective action after AI systems are already in use—this represents a key responsibility for lawmakers to establish clear rules for what happens when AI causes harm.

Different institutions have different roles to play: government agencies and professional societies should focus on funding responsible AI research, states should invest in AI education, and companies should prioritize community involvement and diversity in their development teams.

AI ethics statements and incident databases should receive lower priority compared to other policy options, suggesting that while these transparency measures have value, they are less impactful than concrete auditing and accountability structures.

Resource constraints require strategic choices about which policies to pursue, and this paper provides a framework for institutions to focus their limited budgets on the policies most likely to meaningfully advance responsible AI development rather than attempting to implement all proposed policies equally.