Comment Section for Democratic inputs to AI grant program: lessons learned and implementation plans

This OpenAI blog post discusses the progress of the Democratic Inputs to AI grant program, which aims to involve the public in guiding AI behavior. The program awarded $100,000 to 10 teams out of nearly 1000 applicants from 113 countries to design, build and test ideas that employ democratic methods to determine AI system governance rules. Key insights from the program include the frequent fluctuation of public opinion, the difficulty in bridging digital and cultural divides, the challenge of finding consensus within polarized groups, and the trade-off between consensus and diverse representation. The participants also expressed a mix of hope and concern regarding the future of AI governance. OpenAI aims to address these challenges by forming a "Collective Alignment" team to implement a system for collecting and encoding public input on AI model behavior. They will continue to work with external advisors and grant teams, run pilots, and recruit research engineers to contribute to this initiative. The blog post also provides brief summaries of the projects undertaken by the ten teams awarded grants.