We are seeking a postdoctoral researcher to contribute to an interdisciplinary project focused on enhancing the reasoning and analytical rigor in reports analyzing complex geopolitical situations. This work is part of IARPA's REASON https://www.iarpa.gov/research-programs/reason initiative, under which we've united an extensive international team of computer scientists and subject matter specialists. Our goal is to develop and evaluate systems that aid intelligence analysts, and researchers more broadly, in crafting superior analyses of complex scenarios.
Our goal is to improve the quality of arguments. Are they sound and persuasive? Does the report draw on appropriate evidence? Are there overlooked elements that warrant inclusion? Have potential counterarguments been adequately tackled? This work will involve building higher-level reasoning structures that then call large language models as functions to do the detailed analyses.
Joint work Chris Calison-Burch, Mark Yatskar, Lyle Ungar and a large team.