Having sat through numerous Virtual Roadshows hosted by the Officer Personnel Management (OPM) branch of the Coast Guard’s Personnel Services Center (PSC), I have listened as numerous assignment officers (AOs)and others at OPM discuss the same topics, the same processes, and the same challenges they face.[i] After a while, it begins to sound repetitive, even a bit robotic. During the last one, as the (AOs) touched on the same topics and gave the same brief, a realization slowly dawned on me: The Coast Guard should automate its OPM branch.
For those readers unfamiliar with PSC, and its OPM branch where most of the decisions affecting evaluations, promotions, advancement, career progression, and advanced education occur. OPM specifically handles such issues for the over 8,500 officers and warrant officers on active duty.[ii] To manage this, OPM is split into four divisions: OPM-1, the Boards, Promotions, and Separations Branch; OPM-2, the Officer Assignments Branch; OPM-3, the Officer Evaluations Branch; and OPM-4, the Officer Career Management Branch.
It should be noted before diving into the details that I am not currently assigned to OPM, nor have previously been assigned there. So my perspective is solely as a customer and an outsider, and my arguments are from the perspective of how to improve the process as observed from the outside. Such discussions may already be occurring within OPM, to include discussions of automating various tasks. If not, however, the ideas discussed below provide a path towards such automation.
Taking the Personal Out of Personnel Service Center
I can already hear the protests of some to this idea, but let’s explore what automating OPM would entail, branch by branch. Perhaps the easiest to automate would be OPM-2, the Officer Assignments Branch. With the advances in artificial intelligence and machine learning (AI/ML), coupled with decreasing costs of use, OPM should already be planning a shift to AI/ML for the assignments process.[iii] Developing an algorithm that achieves certain objectives could possibly be low cost and might even exist in a commercial off-the-shelf product. Such a system would ideally be able to calculate several different assignment scenarios in the time it takes the current AOs to complete one full slate (which, as it stands, is a process that lasts from November to March). Any slate produced by AI would still require review by humans, but it would not require the deep dive on every officer that current AOs have to perform now. Parameters could be set to flag certain assignment decisions for review, ensuring the priorities outlined by the Commandant annually in the Guidance to Boards and Panels is met.
I recognize that such a bold shift is inherently uncomfortable for many people, especially those who have risen through the ranks under the current system and are comfortable with the current processes. But as the service continues to experience shortages of officers at the O4 rank, and the likelihood that the Blended Retirement System continues to challenge retention in the future, would it not benefit the service to explore ways to reduce the manning requirements just for the assignments process by leveraging new technology?
Another benefit of automating the assignments process is that it removes the personal aspect from the process. If you have ever been in receipt of orders for an assignment that was nowhere on your e-resume, then you know that it feels personal, as much as OPM-2 tries to convince you otherwise. And the fact that at every Virtual Roadshow the AOs reiterate that the assignments are impersonal highlights the fact that, to many officers who have had their prospects of a dream job dashed, it feels personal, as though the AO was out to stifle your goals.
This continues to be an issue given the frequency in which AOs rotate through accounts. With only a three-year tour, and the constant fleeting up from the junior account (O3 and below) to the senior account (O4 and O5), you are not likely to engage with the same AO more than once, even if you are in a two-year assignment. While OPM likely has a justification for this process, it results in ever-changing interactions with the AOs, and personalizes the assignment process. After all, when Lieutenant Commander Briny tells you this year that your e-resume is unrealistic after her predecessor Lieutenant Commander Salty assigned you your first choice to command an ’87 patrol boat two years ago, it makes the process feel more personal.[iv] But short of extending tour lengths and preventing internal fleet-ups, how could OPM solve this problem?
Enter the cold, impersonal computer.
I know the thought of leaving personnel decisions up to an algorithm in a box likely discomforts many an officer. After all, you cannot leverage your personal relationships, your service reputation, or anything else against the box. All you can do is hope that your performance, professionalism, leadership, and education, as documented in your record throughout your career, put you in the best position possible to get your dream assignment. It approaches a level of objectivity that OPM tries to convey it already has by removing the personal nature. And as a side benefit, it removes what some might argue is a more distasteful aspect of OPM-2: AOs assigning themselves into their next jobs, competing against their peers.
Automating the Future
While OPM-2 arguably handles the here-and-now of the service through officer assignments, OPM-1 handles the future through promotion boards and various panels, as well as the separations of officers through retirement or other means, thus making room for future promotions. Admittedly, I am less familiar with the separations side of OPM-1, having not yet separated myself, but there are likely gains to be had by automating certain processes, such as form review and verification, assuming this isn’t already happening. Instead, let’s focus on how automating the processes for boards and panels could benefit the service.
Currently, to be a part of a board or panel, you must submit your name to OPM-1 to be added to its list. The members of OPM-1 will tell you upfront that they receive more submissions than they have boards and panels, so your prospects of getting selected are apparently rather slim. However, with such a demand signal and so few opportunities, why is it that some members participate in more than one board or panel?
Without any inside information from OPM-1, the explanation likely stems from a few factors. First, OPM, as most units in the service, is probably operating on a limited budget, and thus cannot afford to have numerous officers perform temporary duty, and all its associated travel costs, for every board or panel. This naturally leads to the second likely factor, which is that it pulls officers from inside headquarters for no-cost duty and ease of access. This would all be reasonable enough, and explain the repeated selection of certain officers, if there were only a handful of officers at headquarters and the other units in the D.C. area. But headquarters, and the D.C. area in general, serves as the single largest concentration of officers in the whole service. Given the current practice, which also brings in officers from field units to participate in boards and panels, it begs the question of how some members are afforded multiple opportunities to shape the officer corps through participation in boards and panels.
The reality is that the selection of board and panel members is inherently subjective, based on the preferences of the members of OPM-1, which places immense influence in the hands of a few officers. Automating this process would allow for more randomization in board and panel composition, providing differing and more robust views to the board. It may result in the selection of some officers that others, and OPM in particular, might view as just “average”being selected and thus influencing the future of the officer corps.[v] However, there also are a number of safeguards, such as board deliberations and multiple approvals, that would prevent truly poor officers from being promoted and truly exceptional officers from being passed over.
Some might argue that the majority of officers are just not high performing enough to be allowed to select the future of the officer corps. But such opinions also stifle innovative, and at times disruptive, thinking which could improve the overall quality of the service. After all, the inclusion of a wider pool would likely bring unique insights and experiences that OPM is currently screening out, intentionally or otherwise, by allowing subjectivity into the selection process. Additionally, the effective implementation of automation may allow OPM-1 to automate the actual boards and panels themselves, reducing the required manpower and increasing consistency in the process.
Coding Administrative Efficiency
As an officer, it is not lost on me that I am required to meet stringent deadlines for drafting, submitting, and reviewing Officer Evaluation Reports (OERs) that then take months for OPM-3 to review and validate. I recognize there is a vast difference in scale, with OPM-3 responsible for approximately 10,000 OERs, where I have only a handful. However, OPM-3 is also not drafting these evaluations, but only reviewing them to ensure adherence to policy. As such, this is a prime opportunity for automating some of these processes.
As an outsider, I may not be aware of OPM-3 currently using any type of automation or AI to review and validate OERs. However, if they are using some form of technology, based on the several months it generally takes to receive my validated OER back from OPM-3, it’s likely slow, antiquated, and generally unsupported by the service’s information technology infrastructure.[vi] This is not a knock on the OPM-3 staff, but more a reflection of the service’s challenges with regard to optimizing processes.[vii]
Incorporating this technology could help to streamline the review and validation process, allowing officers to ensure a complete record before boards and panels and decreasing the need for OPM-3 to prioritize review and validation for potential board and panel candidates. Perhaps the most challenging aspect of using AI/ML in the review of OERs is the fact that there are myriad ways supervisors and reporting officers draft their comments, with little standardization provided by OPM-3. However, transitioning to an AI/ML system would enable OPM-3 to set standards for OER comments, improving consistency, and easing the job of board and panel members who have to review the OERs for promotion and selection considerations. Such efforts would put more work on the supervisors and reporting officers to ensure clear, concise, and meaningful commentary.
What Automation Cannot Do: Advice and Counsel
AI and automation can be useful for a number of functions, making decisions based on numerous factors faster than humans in most cases. For all its benefits, however, it has not outpaced the human brain in terms of context.[viii] There is not, at the moment, any system out there than can provide you with tailored career advice for you to consider. This functionality remains an inherently human task, currently performed by OPM-4, the Officer Career Management Branch. Which is richly ironic, since OPM-4 is currently a staff of one officer.[ix]
Part of the issue with OPM-4, historically, is that it was never staffed properly. As a young Lieutenant, I utilized OPM-4’s services and scheduled a career counseling session. At the time, there were no Afloat officers assigned to the office, and so my counseling was rather vague and generalized as the counselor from OPM-4, a Response officer by trade, attempted to answer my questions focused specifically on my unique career path. And the generalities (perform well, balance operational and staff tours, etc.) was generally good advice. But it didn’t help me figure out specific actions I should or should not take to continue advancing my career.
This lack of specificity and small staff resulted in the AOs taking on career counseling responsibilities for members assigned in their accounts. However, given the long assignment cycle, it left just a couple months in the spring each year for the AOs to provide career counseling. As such, it has generally been restricted to those members anticipating transfer in the coming Assignment Year, and OPM-4 responsible for counseling for all others. Understandable, but hardly a recipe for successful career counseling, when officers generally have to identify certain wickets to achieve to be competitive for the next job. Providing that tailored counseling with, at best, four months before the start of the next assignment cycle doesn’t provide officers the opportunity to recognize where they might be coming up short. This process sends the message, unintentionally or otherwise, that you are essentially on your own as an officer, with only the help of your peers and supervisors who may or may not have similar career paths, areas of expertise, and career goals as you.
This presents a question that OPM must figure out before moving forward: Does it want to be in the career management and counseling field? Assuming it does, OPM-4 needs a major overhaul to increase its effectiveness through the addition of staff members representative of the various primary and secondary officer management specialties the service has. In our ever-constant resource-constrained environment, this would normally be a nearly insurmountable challenge, requiring other units to sacrifice billets to enable the required staff growth for OPM-4. However, this can be achieved through other means, namely, automating OPM-1, -2, and -3.
If OPM automates the other branches, as discussed above, it would likely be able to reassign the requisite number of officers to OPM-4 from diverse backgrounds to provide increased specificity and tailored advice to the officer corps, essentially placing the counseling duties the AOs in OPM-2 took on as collateral back in OPM-4 where it is the main mission. It would also remove the personal aspects from the assignment process and instead shift them to the career counseling process, which should be inherently personal. While not everyone would likely be happy with all of the career advice provided by OPM-4, it is likely that the general impression of OPM as an organization would trend more positive.
The Future of OPM
The motto for OPM-1 is “To Be Fair and Equitable.” Personnel management of several thousand officers will continue to present challenges to OPM, to include the appearance of fair and equitable processes. The use of automation can help alleviate some of those challenges, increasing efficacy while providing more objectivity to assignments and promotions. Freeing up resources and manpower could then allow OPM to dedicate more towards the effective counseling and career guidance of an officer corps that is going to face unique challenges in the future as the service and disposition of the officer corps changes. Proactively tackling that problem set now through intelligently planned integration of AI and automation can position OPM to provide superior service for years to come.
Endnotes
[i] Virtual Roadshows are teleconference calls during which specific OPM offices address their various roles and responsibilities, issues such as the assignment process, and answer questions submitted by the officer corps.
[ii] Two other branches of Personnel Services Center, Enlisted Personnel Management (EPM) and Reserve Personnel Management (RPM), handle these issues for the enlisted and reserve workforces, respectively. Personnel Services Command also has a Business Operations Division and a Personnel Services Division, which both handle various issues from resources to uniform policies.
[iii] Thomas Siebel, in his 2019 book Digital Transformation: Survive and Thrive in an Era of Mass Extinction, provides an excellent overview of the status of AI/ML today, as well as Cloud Computing, Big Data, and the Internet of Things.
[iv] Fictional names.
[v] Or “[o]ne of the many high performing officers who form the majority of this grade,” per the Officer Evaluation Report.
[vi] The service is still using Internet Explorer, with all of its security issues, because certain applications haven’t been updated to run on current web browsers. Thus, the likelihood is low that the service is dedicating finite resources to upgrading and maintaining a possible single-use, highly specialized program.
[vii] Some might argue that is exactly why the service should not modernize its processes and systems using AI/ML. However, it is generally more expensive to maintain antiquated and unsupported IT systems than it is to transition to newer platforms. Often, the issue comes with the data migration, and the cost of moving the data between potentially incompatible systems. Incorporating automation in areas where it does not currently exist allows for more flexibility, mitigating the need for data migration between systems, and can decrease support costs.
[viii] Not at mass scale. I’m sure someone will point out some article that proves otherwise. But until that technology is ubiquitously available at low cost while performing or outperforming humans in various capacities, this statement remains true for the majority of human functions.
[ix] OPM-4 experienced a 50% reduction from two officers in the past couple months. So, instead of having 1 career counselor for every 4,200-plus officers, the service now has one counselor for all 8,500-plus officers.
