CAN AI TECHNIQUES HELP INTELLIGENCE ANALYSTS?

Professor Ann E. Nicholson (1)

(1) Monash University, Wellington Rd, Clayton, Victoria 3169. ann.nicholson@monash.edu

 

Many areas of human endeavour require decision making about complex scenarios, where there is: uncertainty about the underlying causal process; limited and possibly inaccurate information about outcomes or contributing factors; and uncertainty about the effects of actions or interventions.

Bayesian networks (BN) have become a state-of-the-art technology to support decision-making under uncertainty. These models can combine data, evidence, opinion and guesstimates to help decisions makers combine probabilities and take into account costs and benefits. The BARD (Bayesian ARgumentation via Delphi) project is funded by IARPA (Intelligence Advanced Research Projects Activity), under the US government’s Office of the Director of National Intelligence, through the CREATE (Crowdsourcing Evidence, Argumentation, Thinking and Evaluation) program.

The aim of the BARD project is to design and build an online tool to support groups of Intelligence Analysts to (i) construct a Bayesian Network that is an argument map of an intelligence scenario and (ii) produce structured and transparent analysis reports, based on that BN. In this keynote address, I will give a brief introduction to Bayesian network technology before presenting an overview of the BARD tool V1.0. I will also discuss both the opportunities and the challenges in such a large, international and interdisciplinary research project.