Come On, Share Your AI Story

“The future is already here – it's just not evenly distributed.”

When I ask someone in the Navy about AI, I typically receive one of two replies.^[i] The first is enthusiastic: "Oh yeah, it's amazing, I just used it to write a recommendation letter for a Sailor." The second reflects uncertainty: "I honestly don't know how it fits into my daily routine, let alone my professional duties." This sharp contrast underscores a significant missed opportunity across the Navy. This divide between enthusiastic adopters and the hesitant majority reveals how much potential remains untapped. What's at stake isn't just administrative efficiency, but also the potential for AI to foster creativity, enhance problem-solving, and inject genuine enjoyment into learning and development.

In her excellent War on the Rocks article “Tackling the Toil,” Alexis Bonnell, CIO at the Air Force Research Laboratory, highlights a problem familiar across the entire defense community: the overwhelming weight of administrative tasks on mission-focused personnel. She writes, "Every minute wasted on unnecessary tasks is a minute stolen from the mission."

The particular genius of Bonnell's formulation lies in its framing of waste as time theft rather than mere inefficiency. This captures service members' frustration who find themselves "bogged down by process"—a euphemism that fails to capture the suffocation of paperwork, tasks, or random, unimportant requests for information—rather than contributing meaningfully to whatever objective they're trying to accomplish. This administrative overload is precisely why some personnel are enthusiastically adopting AI, while others remain buried in the mud.

AI models are not only, as Bonnell notes, tools to ease the toil, but incredible co-workers in creativity.

My own exploration with AI began in 2020 with AI Dungeon, a text-based adventure game backed by early ChatGPT models. Imagine a never-ending choose-your-own-adventure book, or an on-call, 24/7 dungeon master who doesn't need bathroom breaks. In AI Dungeon, a player selects a genre, fantasy for example, creates a character name and is suddenly dropped into a narrative. You type your actions and your dialogue and let your story play out in real-time. It was exciting. It clearly demonstrated AI's potential beyond entertainment. Fast forward to today, and the technological landscape has transformed dramatically—it’s moving weekly, daily in some cases—yet defense adoption still trails behind the available capabilities and education.

More recently, a powerful example of AI's potential emerged when I used an AI tool to create Anki flashcards to help me earn my technician-class radio license. Having always struggled with recalling radio frequency (RF) bands and wanting a deeper appreciation for radio principles, I challenged myself in this area. The interactive experience with the model functioned like having a personal RF tutor, providing an extraordinary learning opportunity. This approach of actively learning through continuous questioning and interaction vividly demonstrated AI's unique educational value.

Another quick example: After attending a Transition Assistance Program (TAP) class, I recognized the need for a tool to aggregate retirement tasks, easily print them, and integrate them into a Google or Microsoft calendar. After a few hours with an LLM, I launched "Navigator," a straightforward web tool addressing this specific challenge. Its simplicity and effectiveness demonstrate how AI can transform administrative burdens into manageable, user-friendly solutions.

Fundamentally, I've learned that I can't simply convince people that my solutions apply to their challenges or persuade them to immediately dive into AI prompting. Instead, what works is connecting with their specific problems and gently offering ways they might overcome their resistance. Suggesting they try different approaches often sparks genuine understanding.

Despite clear benefits, resistance to AI adoption remains. Professor Ethan Mollick, a leading thinker in this area, identifies several reasons behind this hesitation:

These concerns are real. For those of us in the military, you can add classification concerns, OPSEC, and the all-encompassing question bears repeating: "What’s the policy?” All of these warrant thoughtful discussion, yet they shouldn't prevent us from harnessing AI's capacity to address challenges. Moreover, the reluctance to openly share how we use AI limits our collective ability to understand and leverage its full potential. Encouraging transparency and shared learning about AI applications can greatly accelerate its beneficial adoption.

So, ask yourself, how are your peers, seniors, or subordinates using AI? With tools like NIPR GPT available, enabling secure processing of controlled unclassified information (drafting awards or talking points, for example) is incredible, but uneven adoption points to an urgent need for greater awareness and education.

The question has shifted from "Can AI help?" to "How can we collectively leverage AI to tackle our shared challenges?" I believe the answer lies in creating a culture of shared learning and transparency around AI use cases. We should all be sharing problems we've identified in our workplaces, the administrative toil we face, and how we've used AI models to solve these challenges. From creative approaches to studying, teaching, wargaming, translating, coding, and military planning—the applications are innumerable.

As Bonnell concludes, "in an age where time is as critical as any advanced weapon or capability, the ability and courage to eliminate waste is nothing short of a strategic necessity."  If we genuinely value the time and talent of our defense workforce, integrating LLMs to reduce administrative toil is not merely beneficial, it's strategically essential for national security.

Our "get-stuff-done" spirit is already at work, with many quietly solving these challenges. Leadership at every level should encourage transparency and the sharing of best practices. Imagine a simple repository that catalogues our challenges and individual solutions for the services. Or a library of prompts or AI approaches that tackles some knowledge work inherent in the DoD. Such a platform would accelerate adoption, prevent redundant efforts, and create a multiplier effect across the service.

By embracing this technology and openly exchanging experiences, we can transform burdens into solutions—enhancing our focus, efficiency, and maybe, just maybe, making us better tomorrow than we were today.

CDR Chrisotpher Nelson is a naval intelligence officer. The opinions here are his own and are not necessarily reflective of the U.S. Navy or Department of Defense.

Note:

^[i]Yes, AI and Large Language Models are not the same, but in this essay, I mostly focus on LLMs and often use "AI" to refer to them. I know AI includes many other technologies beyond LLMs, but I use the term "AI" because it’s widely understood. As different AI tools—like computer vision, reinforcement learning, and autonomous systems—are becoming more connected, the lines between them are fading. So, the points I make here about LLMs often apply to the broader world of AI as well.