Experts Advocate for Evidence-Based AI Policy as U.S. Unveils Strategic Action Plan
Top scholars and U.S. government unveil new strategies advocating evidence-based AI policy for effective governance.
Key Points
- • Top scholars call for evidence-based AI policy to inform governance.
- • U.S. unveils 'Winning the Race: America’s AI Action Plan' with over 90 recommendations.
- • Two initiatives emphasize the need for transparency and collaboration in AI development.
- • Challenges remain in defining credible evidence and fostering consensus in AI policy.
A recent call from top scholars for an evidence-based approach to AI policy coincides with the U.S. government's introduction of its comprehensive AI action plan, highlighting the urgent need for structured governance in this rapidly evolving field.
On July 31, 2025, a significant paper published in *Science* articulated the necessity for policymakers to leverage scientific research in developing AI governance strategies. This initiative was driven by 20 scholars from Stanford’s Institute for Human-Centered AI (HAI), among them Fei-Fei Li and Rishi Bommasani, who urged for credible evidence and transparency from AI companies when forming regulations. Bommasani explained that while evidence-backed policies are recognized as vital, their implementation is complex in the dynamic AI landscape. The scholars pointed out ongoing challenges, such as determining what qualifies as credible evidence and fostering environments that encourage consensus among stakeholders in the AI community.
In a parallel movement, the White House unveiled 'Winning the Race: America’s AI Action Plan' on July 23, 2025, targeting U.S. leadership in AI technology. This action plan includes over 90 policy recommendations spread across three pillars: Accelerating AI Innovation, Building American AI Infrastructure, and Leading in International AI Diplomacy and Security. It emphasizes actions to minimize federal regulatory barriers faced by AI innovation while promoting open-source models and necessitating regulatory sandboxes for practical testing. In addition, it stresses the importance of investing in infrastructure to support AI advancements and protecting against foreign challenges, particularly from China.
“Policymakers must create environments that support the generation and utilization of high-quality evidence in shaping AI policy,” asserted the scholarly paper. This sentiment echoes the overarching themes of the U.S. action plan, which advocates for reduced regulation, streamlined data center permitting, and collaboration with allies to maintain global AI leadership. Both initiatives reflect a marked shift towards a more evidence-based governance approach in AI, aiming to address the pressing need for regulatory clarity and effectiveness amidst a rapidly growing technological landscape.