Why the Department of Defense Should Create an AI Red Team

What is adversarial machine learning (AML)? AML is the purposeful manipulation of data or code to cause a machine learning (ML) algorithm to misfunction or present false predictions. A popular example of AML is from a team at Google that carried out an experiment on GoogLeNet, a convolutional neural network architecture that won the ImageNet Large Scale Visual Recognition Challenge in 2014. Adding noise to an image of a panda and digitally changing its characteristic led the program to more highly predict that the image was a gibbon. This type of manipulation is relatively easy to execute with just a few bits of code inserted into the original algorithm.

Read more

War is a Learning Competition: How a Culture of Debrief Can Improve Multi-Domain Operations

The Multi-Domain Operations (MDO) community continues to evolve and progress. MDO is, and will be the fundamental enabler for Joint All-Domain Command and Control (JADC2) and the way our nation fights future wars. As the maturing community integrates new concepts and processes, Multi-Domain Operators must identify and engrain the valuable lessons along the way. Creating a set of standards to capture feedback and drive improvement is vital for development in any organization. The debrief culture of the US Air Force fighter community, among others, is well known for its direct, highly effective feedback and learning methods. This type of focused feedback is important to the fighter community because the debrief is where the majority of learning takes place. The MDO community would benefit greatly by utilizing this debrief culture as a model from which to develop its own unique culture of consistent, iterative improvement. Because a standard day, or sortie-equivalent, is not yet fully fleshed out for Multi-Domain Operators, the purpose of this paper is to convey the necessity for debriefing lessons learned, and provide best practices in their current form. The ultimate objective is to create a foundation for the MDO community to adapt these practices as the details and nuance of its daily execution become more specific and clear.

Read more