Air Force-MIT Phantoms PREP to Buy and Use Artificial Intelligence
By Air Force Captain James W. Hanley
EDITOR'S NOTE: Across the Defense Department, programs are springing up that give contracting professionals opportunities to directly experience the challenges the current contracting process presents to providers of advanced technology and services. These chances to get down and dirty in emerging technology and companies supplying it are invaluable to participants by all accounts. In the January 2021 issue of Contract Management, Air Force Technical Sergeant James Pitcher called his participation in the Defense Ventures venture capital industry immersion program a “transformational experience.” The Defense Innovation Unit-Defense Acquisition University Immersive Commercial Acquisition Program announced its first cohort last month in Chicago. Here’s hoping these DoD programs grow and proliferate to give civilian and military contracting professionals even more opportunities to experience the innovative technologies and companies they need to know. And let’s start seeing the same forward-looking, business-acumen-building activities in civilian agencies as well!
In November 2021, I was fully immersed in the final preparation for my upcoming unlimited contracting warrant board exam. For the past several weeks, I had been balancing work, study, and several side projects that were beginning to come to fruition. Oddly enough, it was one of those side projects that led to the most amazing professional opportunity in my career.
My first opportunity to engage with the artificial intelligence ecosystem came through an email to the Hanscom Air Force Base Innovation Team. It sought a volunteer to help address the best ways for the Department of Defense to acquire artificial intelligence (AI) systems.
At the time, I had no idea what AI was, but I figured it would be a great learning experience and an opportunity to shape some low-level policy. I worked with an exceptional senior leader in the engineering community and produced a white paper highlighting functional areas that needed to be addressed for the Air Force to effectively deliver AI capabilities to the warfighter.
Shortly thereafter, a program manager friend reached out to let me know he was working on AI acquisition policy for an intriguing organization I had never heard of: The Air Force-Massachusetts Institute of Technology (MIT) AI Accelerator (AIA).
We began sharing notes and discussing the crossover between the problems he was trying to address, and the work I had recently completed with the innovation team. Within a few weeks, the AIA issued a call for applications for their next – and largest, with 16 members – cohort of Phantom Fellows.
For the first time in the program’s short history, the call also asked for a contracting officer. Just days after I successfully passed my unlimited warrant board exam, I found out I had been selected as the first contracting officer for the Phantom Fellowship Program.
The Department of the Air Force MIT AIA was established in 2019 under a five-year cooperative agreement for $15 million per year, or $75 million overall. Tapping MIT expertise and resources, the AIA works to advance the underlying science of AI and facilitate societal applications via fundamental research and education.
The AIA thrives by assembling interdisciplinary project teams with members from MIT, MIT Lincoln Lab, and active and reserve military personnel. The Phantom program was established in 2021 and has expanded to a four-month fellowship that allows members from across the Air Force to learn about AI and shape cutting-edge research and policy.
After the shock of my selection to the Phantom program wore off, I considered how I could bring value as the sole contracting professional. First, I had to figure out AIA’s value proposition. I’ve come to understand the AIA is a multi-faceted organization with two main missions: research and education.
The AIA is charged with conducting fundamental research to advance the state of the art for AI. By leveraging both the world-renowned expertise and the resources of MIT and its Lincoln Lab, the AIA facilitates public and defense use cases.
Interdisciplinary teams are built around each research project including representation from Air Force members, MIT staff, and Lincoln Lab personnel. This organizational structure allows the AIA to test novel technical approaches on real Air Force operational issues.
Each research project runs independently, which allows stakeholders at different command levels to get involved where they may be most comfortable. Some projects only need operational data, while others need transition partners for prototype applications. Regardless of the organic technical prowess of an organization, there is likely a way to benefit from or advance the research being done at the AIA.
Being the only Air Force unit specifically dedicated to the advancement and proliferation of AI, the AIA takes on the more classic mission to educate, train, and equip the force to understand and apply AI at all levels. This is a simple but arduous task given the nuance and heavy debate about the field of AI. It becomes exponentially more difficult when you factor in the different ways that servicemembers will interact with AI.
Useful educational tools must meet people at their level of application and understanding. How can a unit of roughly 15 permanent members satisfy this need? The simple answer is that the AIA cannot meet that mark without help, and that is what the Phantom Fellowship provides.
The fellowship is a four-month remote program with several in-person stretches of temporary duty for organizational integration. The program currently targets engineers, program managers, and contracting officers and is open to any Air Force Specialty Code (AFSC) on the enlisted side (given technical expertise), while specifying the inclusion of at least one public affairs member.
This combination of functional designations may not seem to make much sense at first but there is a method to the madness. Phantoms are brought in to augment research teams and help with education and policy efforts taking place on the Air Force side of the AIA. The acronym “PREP” encapsulates the work the AIA accomplishes and illuminates the value of the Phantom program.
P – Partnerships: In operational workstreams (for example, the prototyping of advanced anomaly detection), we have worked with functional communities such as operations research, and operational testing and evaluation. We also worked with organizations such as the Morpheus team under the chief of staff of the Air Force’s Strategic Studies Group to establish new force development measures to address talent management gaps.
R – Research: Even though I am far from a proficient coder, I still was able to aid in data discovery and exploration research the AIA is conducting. I was able to lay out the acquisition process in a way that would allow the research team to hone their capabilities to address real issues practitioners face every day. We focused on data query challenges in conducting robust market research, and highlighted ways to unlock the potential in using acquisition data from past requirements to enable better acquisition planning.
E – Education: We have established agreements with the open learning department of MIT, allowing us to host their content on Non-Classified Internet Protocol Router Network (NIPR) environments, unlocking world-class education resources for all of the Department of Defense.
P – Policy: I’ve had the privilege of co-authoring several whitepapers and proposals that have been routed through to the Office of the Secretary of the Air Force. The papers address potential acquisition policy blind spots on data and AI implications. The most glaring of them is the lack of contracting language or clauses requiring contractors to provide clean, properly formatted data usable for machine learning beyond a particular contract. Our current acquisition environment is not conducive to creating an interconnected, AI-ready data pipeline, but we are laying the groundwork to address this and similar shortcomings.
Looking back on the last four months, it’s mind-blowing and deeply humbling to realize the impact ordinary Airmen and Guardians can have when given the proper resources, autonomy, and senior leader support.
My time getting to know my cohort of Phantoms made it abundantly clear there is a growing contingent of outstanding coders and programmers in our enlisted force who absolutely need to be empowered to use their technical skills to improve the force from within. Unfortunately, those skills do not always align with their AFSC. Nevertheless, there is an enormous well of untapped potential we cannot continue to ignore.
Lastly, it is important to highlight why this type of work and this program are so applicable to the contracting community. Though it took me some time to fully appreciate the connection, it’s now apparent that these basic tenets should become a part of the acquisition curriculum.
Simply put, AI problems are data problems. The acquisition community deals with and creates an enormous amount of data, sometimes so much we do not even use it properly. We routinely find ourselves searching for the proverbial needle in a haystack when we are trying to accomplish simple tasks.
Beyond that, we make momentous decisions every day with taxpayer funds, but can we say we are using historical data and trends most effectively when doing so? If we were honest the answer would be "No." There are both infrastructure and application roadblocks, but without the proper understanding of the importance of data and the state of the possible, we cannot begin to address longstanding problems using the proper modern solutions.
The insights I have gained through the Phantom program are so deeply valuable and essential that I doubt I will ever look at contracting or acquisitions quite the same again. I have learned so much from both the technical and acquisition perspective that is truly helping me grow as a contracting officer and as a leader.
For these reasons I cannot think of a better program for contracting professionals who are looking to make real and lasting change in the federal acquisition arena.
Captain James W. Hanley holds an unlimited warrant as a contracting officer. He was formerly assigned as the deputy procuring contracting officer for the CloudOne Program, supporting the Air Force cloud hosting. He recently transitioned to be the contracting advisor to the strategic innovation team at the Air Force Kessel Run software factory in Boston, Massachusetts.