The Institute for Security and Technology is excited to announce a partnership with Metaculus, a community forecasting platform designed to generate ML-optimized collective intelligence on topics of global importance. Optimal aggregation of quantitative probabilistic forecasts is at the core of what Metaculus produces: remarkably accurate predictions of the probabilities of future events.
This partnership will build on IST’s existing work on AI safety and nuclear confidence-building measures. Today, IST and Metaculus release Forecasting the AI and Nuclear Landscape. During the initial phase of this initiative, the partners are focusing on useful interventions in the United States-China relationship. This project draws on nuclear and policy subject matter experts to inform professional forecasters’ predictions within the following categories: AI-Nuclear Integration, Measures of US-China Tensions, and Nuclear Use.
“We’re thrilled to work with Metaculus on an evidence-based, forward-thinking approach to AI safety and nuclear CBMs. We look forward to applying collective intelligence to our quest to further mitigate nuclear risk and enhance the security of AI,” said Chief Executive Officer Philip Reiner.
Metaculus’s approach to collective intelligence stands out as a unique example of innovation in the field of epistemic infrastructure. Its forecasts are transparent, easily interpretable, and collaborative.
“We are honored to work with IST to better understand the future of two of the most critical technologies of our time. IST’s deep national security experience combines well with our empirically grounded forecasting methods to produce novel insights within a complex environment,” said Metaculus Chief Executive Officer Gaia Dempsey.