Defense Science Board 60th Anniversary

Defense Science Board 60th Anniversary.

On September 20, 2016, the Defense Science Board celebrated its 60th anniversary with an all-day event in Washington DC focused on a variety of topics from a host of speakers and panelists. Dr. Popp took part on a panel focused on Artificial Intelligence (AI): What’s Real, What’s not and is this the DoD Third Offset?

The panel was moderated by the honorable Zachary Lemnios of IBM, and the other panelists included: Dr. Manuela Veloso from CMU, Dr. Bill Mark from SRI International, and Dr. David Kenny from IBM Watson. The agenda for the event can be found here: DSB agenda

As background for the panel, the Department of Defense has made significant investments over the past 50 years, in what many regard as three waves of Artificial Intelligence. The first wave (1950-1970) launched the academic field of computer science, opened an era of discovery and set the foundation for signal processing, computer vision, computer speech and language understanding. The second wave (1970-1990) saw codification of knowledge in expert systems, using rule bases, and beginnings of simple machine inference to do reasoning (think things like computer chess), along with exploration of computer architectures, specialized for AI applications. The third wave (1990- present) launched the era of large scale robotics, including autonomous machines, along with real breakthroughs in the use of neural network architectures, inspired by better understanding of how the brain works. The panel focus was to shed light on how Artificial Intelligence has been adopted in the commercial sector, what new value and impact has resulted, the key remaining technical barriers to adopting AI in the defense sector and how AI could be positioned for an enduring strategic national security advantage.

Dr. Popp’s remarks addressed AI from an “analytic intelligence” perspective to help clients understand people and their behaviors on critical, complex decision-making problems. He discussed the importance of utilizing various analytic methods (including qualitative, quantitative, and mix-method approaches) and multidisciplinary social science techniques (from economics, political science, anthropology, sociology, psychology, social psychology, etc) to help clients address the complexity, ambiguity and uncertainty inherent in understanding people and their behaviors. He indicated the objective for analytic intelligence is to help clients make more informed and better decisions in understanding people and their behaviors by providing them with deeper analyses and clarifying insights.

He also discussed five critical barriers to adopting this type of technology, namely, automated coding – using automated methods to code largely text-based data into “soft” metrics for analysis such as respect, honor, dignity, intent, grievance, motivation, influence, and perceptions; machine translation – a lot of data germane for human behavior problems in the social sciences are in foreign languages, and important meaning can oftentimes get lost in the translation; analytic integration – integrating multiple results and findings representing multiple academic and professional disciplines and analytic approaches into a comprehensive and robust analysis relevant to the operational community or decision-maker; effective presentation – distilling highly technical information and presenting it with parsimony and precision in a manner that is accessible and useful to a broad community of interest; and augmented intelligence – machines replicating the critical thinking and analytic reasoning skills and experiences that humans have.