Steve Escaravage
Senior Vice President of Artificial Intelligence,Booz Allen Hamilton
Innovations in AI technology can improve national security, but a knowledgeable workforce is vital for these tools to be truly effective.
A virtual dogfight between an Air Force F-16 pilot and an artificial intelligence (AI) program in August resulted in a resounding win. The AI “pilot” beat a highly trained fighter pilot in each of five head-to-head scenarios. Although this was a small-scale test in a virtual simulated environment, the results highlight how people can work with AI to leap forward in capabilities. The Pentagon is now clamoring for a real-world dogfight, which Secretary of Defense Mark Esper says will take place using real fighter planes in 2024.
AI is now one of the most significant tools available to us, with the power to bolster our national defense, especially as present-day competitors — nations like China and Russia — pour massive investments into their own militaries. With massive quantities of information and data at play, the only way to make sense of it all in a short amount of time is by harnessing artificial intelligence’s processing power.
AI can quickly sift through mountains of data, enabling analysts to proactively identify security threats while saving time for better and more well-informed decisions. These programs can help unearth terror networks, drug cartels, and foreign interference in our elections. They can also monitor our nation’s critical equipment and infrastructure to ensure it’s working and ready to go the moment our military, law enforcement, and first responders need it. Leveraging AI enables people to accomplish in minutes what would normally take weeks or months.
Unfortunately, AI’s tremendous potential comes with some technical limitations that make implementation difficult.The biggest predictor for successful adoption of artificial intelligence — for national security as well as commercial endeavors — isn’t access to data, large amounts of computing power, or exquisitely designed computer models: it’s the workforce. Not everyone needs to play the role of a data scientist, but we should all be involved in understanding AI’s fundamentals, its limitations, and how this technology can be used across a wide variety of opportunities.
An AI-trained workforce understands there is no plug-and-play version of AI. it must be built and tailored to each organization’s specific needs. Problems must be narrowly defined. Solutions are developed through small rapid improvements, not by a single elegant solution. Using AI requires time to train, test, and refine for the best results.
For example, consider an intelligence analyst who can review twenty satellite pictures per day with a 95 percent chance of producing error-free results. Moving to an AI-enabled solution might result in the processing of one thousand pictures per day, but with an initial accuracy rate of only 45 percent. The algorithm, however, can quickly improve as more training occurs, ultimately freeing the analyst to do more creative and critical thinking while the AI handles the mundane, monotonous task loads.
Federal, state, and local governments will continue to explore and adopt AI to better serve its citizens. Widespread adoption of this technology, however, requires knowledge and experience to ensure solutions are integrated correctly, to anticipate challenges in real-world use, and to provide the large-scale workforce training so critical to overall success.
The extraordinary leaps that have taken place with software, cloud computing, and streaming services are in store for us again as we increasingly harness the power of large datasets, computing power, and proven AI models in support of our national security.
America will increasingly rely on AI’s transformative power to maintain security and provide for our nation’s defense. Reaching our fullest potential, however, requires one critical component: a knowledgeable workforce that knows how to partner people with AI to produce meaningful outcomes. Only then can we use the full capabilities of AI to help solve our toughest challenges.
