$20K in bounties for AI safety public materials

Abstract

The Effective Altruism Forum announces a $20,000 reward for the most accessible public explanations of AI safety concepts. Recognizing that few among technologists, ML researchers, and policymakers are seriously contemplating AI existential safety, the competition aims to increase awareness and understanding of AI safety and risk. The authors stress that there’s a lack of strong introductory resources to AI safety for those without a background in machine learning or AI-risk, contributing to the lack of serious consideration. The initiative encourages distilling existing research and ideas into a cohesive and understandable format for broad audiences, and focuses on various categories of work such as executive summaries, and explainers of relevant AI safety topics like weaponization of AI, power-seeking behavior, and “enfeeblement problem”, among others. The competition runs until December 31, 2022, and winners may receive up to $2,000 each. – AI-generated abstract.