Evolving Artificial Intelligence Continues to Challenge Defense Strategies

Intelligence resources of all kinds contribute to changing landscape of defense.

he future is opening an array of ideas that are melding the use of Artificial Intelligence with defense in ways that challenge classical norms. The future will take advantage and make use of more publicly available open-source data to create a battle plan that is more efficient and less predictable. The fire hose of incoming data is changing the traditional practices employed and creating new opportunities. As threats are no longer constrained by geography, the internet has enabled attacks through cyberspace. Protected borders and geography have provided the United States with security in the past, but as we become more connected this is no longer the case. Today we can be attacked without a shot being fired from known and un-known advisories. The use of power is not what it used to be. Of course, wars fought with beans and bullets will continue, but the threat has expanded to include a whole other level of infrastructure and civil unrest. The United States as a leader in connectivity is both powerful and vulnerable at the same time. This has resulted in defense strategies that we have to under-stand and anticipate with a whole new array of threats. Beyond this the need to look beyond those that we believe to be our primary adviso-ries, we must look to those that are non-state actors or domestic groups wishing to cause harm. The ability to configure data in helpful data sets continues to challenge the most advanced minds. As open-source information once thought to be ancillary to the primary information from front line sensors, is gaining greater prominence. Today the merge of data from social media and crowdsourcing has contributed significantly to identifying those that attacked the Capitol on January 6th as an example. It is the opinion of many that this trend will continue with many front-line sensors, being validat-ed through the use of back-end data found in non-defense data centers.

The abundance of data combined into the building blocks of our defense systems has challenged the most advance high-performance computer systems. The architecture of Edge computing has created a network of pre-processing that creates efficiencies, but also has increased the complexity and subsequently the possibility for risk through the conveyance of eroneous data. The use of a public, social media source like Facebook to contribute to the identification of a threat, in support of an airstrike creates several questions regarding the quality of the data. The speed of decisions continues to challenge us and increases the possibility of error. In 1962 during the Cuban missile crisis, the President had days of deliberation after reviewing the images from the U2 spy plane. After 9/11, President Bush had just a few hours to interpret the data to determine responsibility and how the United States would react. The need for faster and fast-er decision-making is a must. The simple fact is that by increasing the speed we can outstrip the enemies’ communications and soften their defenses through the element of surprise. The ability to combine pre-processing at the Edge and open-source data into actionable intelligence is key to our future defense matrix. The topology of data collection and process-ing will be the kin to the Arms Race of the last 50 years. The future warfare tactics will be determined in the data center by those that command the fastest processors and networks. As the amount of data is doubling every 24 months, to be able to tee up data sets through Edge solutions, combine it would advance algorithms that support the latest in AI techniques, and convey actionable intelligence to the field will be key. By creating battleplans that defy our playbook with asymmetrical, surprise responses will leave our advisories dazed and confused. In Japan, they refer to it as “Kuzushi” or using your opponent’s position of being off-balance or ex-posed against them. The ability to determine threats on a human level has been achieved through the “gift of fear”. When troop movement is witnessed, it has been the human intelligence that has determined whether the threat is advancing or retreating. The use of AI will augment this calculation to determine the greater implications. In history, there are numerous examples of miscalculations based on human intelligence. The one that comes to mind is Hitler’s belief that Normandy was a bluff, and the real attack would come at the Porta Calais based on intelligence gathered from a double agent code name Garbo, Juan Pujol Garcia. Now think of automating the tasks through advance artificial intelligence. Think about a solution that removes the human intelligence that takes time and is mundane and replaces it with automation that paints a picture from many seemingly irrelevant “pixels” to form a more comprehensive threat analysis. A form of this can be witnessed in the Israeli Iron dome that takes immediate action in thwarting incoming missiles. The near term will not be a fully automated system based on AI, it will be open-source information and AI, used in conjunction with known strategies to scrub and challenge the human assumptions regarding the proper course of action. Using more data points, decisions will be made with a richer competition of ideas that theoretically should lead to a better battle plan. The biggest obstacle in implementing an advanced AI-based solution is culture. The entrenched response or playbook is not only known by our advisories but also etched into the fabric of our military. Information is covet-ed and the mantra of “Need to know” is pervasive. We must be open to integration driven by JADC2 from the AFRL. We must also recognize that much of the intelligence used in our sensor to shooter strategy, may be achieved using commercial, non-defense specific sources. The strategy employed must be open to sourcing data in unconventional ways from commercial sources already in position. From weather sat-ellites to Social Media platforms, from crowd-sourcing to GPS tracking, unorthodox solutions will prove to be the best edge over the enemy. An example of intelligence that came from the public occurred last July when a satellite image picked up a fire in Iran that look random, but as the image reach social media within hours it was determined that it was an explosion at a nuclear facility housing enriched uranium. By early that evening the Israeli government was being asked what if any responsibility they may have had in the sabotage. Through speed and accuracy, we will be able to form a much more complete picture of the threat and the possible asymmetrical responses available. This sounds like Science Fiction, and costs will indeed be high, but in the end, it will be AI that augments human intelligence that op-timizes our response. We are spending billions to integrate systems across the branches of the service, with much of the data we seek already available through open sources. This enables us to remove much of the calculus of money spent to achieve primary data, to tapping into data that already exists. It will also change the con-versation from either/or – either human-based decisions or AI based, to an augmentation that gives clarity to the facts.

 

Leave a Reply

Your email address will not be published. Required fields are marked *