Eva VivaltBoard of Directors member
Research institute · Research orgs · Delaware · Est. 2022
Forecasting Research Institute (FRI) is a Delaware-registered 501(c)(3) nonprofit that develops and applies forecasting methods to support decision-making on high-stakes issues. It builds on the forecasting work of Chief Scientist Philip Tetlock by running large-scale expert forecasting studies on topics including artificial intelligence (AI) progress and catastrophic risk, biosecurity, and geopolitical risks, and then translating results for policymakers, researchers, and the public.
No curated programs yet. Check the careers page ↗.
Introduction In the second post in our series on the accuracy of AI progress forecasts, we look at a few ways we are aiming to improve the accuracy and relevance of LEAP forecasts in the future. In the previous post we looked at AI progress forecasts from across FRI’s work. Broad
How Accurate Have AI Progress Forecasts Been So Far?We take an initial look at where median forecasts are (mostly) underestimating and (sometimes) overestimating AI progress, adoption, and impact.
Automated AI Risk Outlook (AIRO)A regularly updated dashboard that tracks the probability of catastrophic risk events, as forecast by an ensemble of frontier AI models.
Forecasting the Impacts of ASL-3 Safeguards on Biosecurity RisksA study to better understand the effectiveness of Anthropic's AI Safety Level 3 measures at reducing biosecurity risks.
Forecasting AI Cyber Risks and Capabilities: Results of a 2025 Pilot StudyA pilot study investigating how AI capabilities may affect near-term cybersecurity risk, focusing on two high-impact cyberattack pathways: data-damaging worm attacks and cyberattacks against the U.S. electrical grid.
Measuring Judgment Quality in Natural-Language Explanations: Evidence from Forecasting TournamentsWe introduce EQMs, a scalable, interpretable method for extracting judgment-relevant information from written rationales.
Forecastbench A Dynamic Benchmark Of Ai Forecasting CapabilitiesRoots Of Disagreement On Ai RiskAi Conditional TreesNear Term Xpt AccuracyExistential Risk Persuasion TournamentLongitudinal Expert Ai Panel LeapFRI published a technical report forecasting the impacts of Anthropic’s AI Safety Level 3 (ASL-3) safeguards on biosecurity risks, based on survey forecasts about expected financial damages from large-scale human-caused outbreaks under multiple scenarios.
Forecasting AI Cyber Risks and Capabilities: Results of a 2025 Pilot Study publishedFRI published a working paper presenting results of a 2025 pilot study forecasting how AI capabilities may affect near-term cybersecurity risk, focusing on data-damaging worm attacks and cyberattacks against the U.S. electrical grid.
Forecasting the Economic Effects of AI revisedThe same working paper page lists a revision date following the initial publication date.
Forecasting Existential Risks: Evidence from a Long-Run Forecasting Tournament (XPT) published/revisedFRI published a paper documenting results from the Existential Risk Persuasion Tournament, including participation numbers and observed disagreement/convergence patterns.
Forecasting Proficiency Test: A General Use Assessment of Forecasting Ability (published)FRI published a research piece describing a general-use assessment approach for forecasting proficiency for decision-makers relying on probabilistic predictions.
Longitudinal Expert AI Panel (LEAP) project overview (updated)FRI’s LEAP project description states that LEAP is a three-year project collecting monthly forecasts about AI capabilities, adoption, and impacts and describes the lead institutions and overall structure.