Using REASON to Improve Intelligence Analysis

Every day, the Intelligence Community (IC) labors to provide policymakers with insightful analysis to answer and inform on some of the toughest national security questions and issues. More often than not, the IC is able to provide ground-truth on seemingly murky problems that may otherwise not be discovered or explained.

However, even with the IC’s vast resources and expertise, analysts may still not have the full range of available information to counter arguments and draw the most robust conclusions. This can potentially have devastating consequences if policymakers make the wrong decision based on the intelligence they are provided. 

For example, the Commission on the Intelligence Capabilities of the United States regarding Weapons of Mass Destruction noted in their Unclassified “Report to the President of the United States” dated March 31, 2005, that, “virtually all of the Intelligence Community’s information on Iraq’s alleged mobile biological weapons facilities was supplied by a source, codenamed ‘Curveball,’ who was a fabricator.” Contrary evidence was discounted or ignored, and evidence of Curveball’s fabrications was not passed on to policymakers.

Avoiding similar outcomes of unintentionally providing erroneous information is a constant challenge for the IC. To address this challenge, IARPA has developed the Rapid Explanation, Analysis and Sourcing Online (REASON) program. REASON’s objective is to develop novel technologies to help intelligence analysts substantially improve evidence and reasoning in draft analytic reports.

REASON will assist and enhance analysts’ work by pointing them to previously unconsidered key pieces of evidence and by helping them determine which alternative explanations have the strongest support. To achieve this, REASON will exploit recent advances in artificial intelligence (AI) so evidence is provided automatically and on demand as the analyst works on a report. REASON won’t perform the analysis or write the report for the analyst, but it will seek to strengthen the report’s conclusions.

“IC analysts have a really difficult job, which requires them to sort through huge amounts of often uncertain and conflicting information as they strive to answer intelligence questions,” said REASON Program Manager, Dr. Steven Rieber. “I believe REASON will help make an analyst’s job easier and increase the value of the available information.”

Dr. Rieber cited the lead-up to Russia invading Ukraine as an example of the impact of high-quality intelligence. “Policymakers said IC analysts did an amazing job updating them with fast-breaking information about Russia’s intentions and movements,” Dr. Rieber said. 

When launched, IARPA will invest nearly four years into developing REASON’s underlying AI technology. To do this, IARPA will select performer teams to conduct research and development to build AI systems and then an independent testing and evaluation (T&E) team will evaluate the systems. T&E will ensure that REASON AI technology is effective in helping analysts discover valuable evidence, identify strengths and weaknesses in reasoning, and produce higher-quality reports. 

While REASON is just getting started and there’s no guarantee of success, Dr. Rieber said he is optimistic. “I think AI is far enough advanced that REASON stands a very good chance of becoming a reality.”

Read more at IARPA