Artificial Intelligence Governance

Abstract

Research on AI governance is important towards potentially useful and constraining affordable misuse, reduce new risks and economic trends that threaten to disrupt public political and economic trends, and drive off target as interest in advanced AI systems and the norms, focal points, and use of new AI research are potentially transformative and governance institutions aim to prevent. Potential public benefits from policy community re-using AI research are enormous, including reduced economic instability. A fundamental challenge in AI governance is a cognitive framing challenge: governing AI research requires understanding new kinds of safety risks, performance goals, and intended applications that advanced AI systems will make possible. Specifically, the letter focuses on how AI research could mitigate issues such as the possibility of AI capabilities getting concentrated within a small and hard-to-regulate group of actors, and ultimately recommends the prioritization of open research and collaboration, with concern for long-term social and economic looming to the forefront of coalitions if AI becomes an increasingly important aspect of the future economy and society.