[Paper Review] Smart Policies for Artificial Intelligence

Miles Brundage, Joanna J. Bryson|arXiv (Cornell University)|Aug 29, 2016

Big Data and Business Intelligence20 citations

TL;DR

This paper argues that artificial intelligence is already governed by a fragmented, de facto policy landscape and calls for more informed, integrated, and forward-looking AI policy. Drawing on lessons from other scientific fields, it proposes structured reforms to make AI governance more anticipatory, effective, and coherent.

ABSTRACT

We argue that there already exists de facto artificial intelligence policy - a patchwork of policies impacting the field of AI's development in myriad ways. The key question related to AI policy, then, is not whether AI should be governed at all, but how it is currently being governed, and how that governance might become more informed, integrated, effective, and anticipatory. We describe the main components of de facto AI policy and make some recommendations for how AI policy can be improved, drawing on lessons from other scientific and technological domains.

Motivation & Objective

• To analyze the current state of de facto AI policy, which is fragmented and reactive rather than strategic.
• To identify gaps and inefficiencies in existing governance mechanisms for AI development.
• To propose a more systematic, forward-looking approach to AI policy inspired by successful models in other scientific and technological domains.
• To encourage policymakers and researchers to proactively shape AI development through coordinated, evidence-based strategies.
• To advocate for integrating ethical, technical, and societal considerations into a unified policy framework for AI.

Proposed method

• Analyzing existing policy mechanisms affecting AI development across government, industry, and academia.
• Drawing analogies from policy frameworks in biotechnology, nuclear energy, and environmental science to inform AI governance.
• Identifying core components of de facto AI policy, including funding allocation, research ethics, and regulatory oversight.
• Recommending the creation of interdisciplinary policy councils to coordinate long-term AI planning.
• Proposing the adoption of foresight methodologies such as scenario planning and impact assessments in AI policy development.
• Emphasizing the need for transparency, public engagement, and adaptive governance in AI policy design.

Experimental results

Research questions

• RQ1 What are the current, informal mechanisms shaping AI development, and how do they function as de facto policy?
• RQ2 How can AI policy be made more anticipatory and integrated, rather than reactive and fragmented?
• RQ3 What lessons can be drawn from policy approaches in other scientific and technological fields for governing AI?
• RQ4 What institutional structures are needed to ensure coherent, long-term AI governance?
• RQ5 How can ethical, technical, and societal dimensions be better integrated into AI policy frameworks?

Key findings

• A de facto AI policy already exists through a patchwork of regulations, funding decisions, and institutional practices, though it lacks coherence.
• Current governance mechanisms are often reactive, fragmented, and insufficiently coordinated across sectors and jurisdictions.
• Successful policy models from fields like biotechnology and nuclear energy offer transferable strategies for AI governance.
• There is a critical need for anticipatory governance that includes foresight tools such as scenario planning and impact assessments.
• Integrating diverse stakeholders—including researchers, ethicists, and the public—into policy design improves legitimacy and effectiveness.
• The paper concludes that the key challenge is not whether AI should be governed, but how to improve existing governance to be more informed, integrated, and forward-looking.