Battle Research Group (BRG) is a nonprofit research organization dedicated to improving collective understanding of modern war. We apply qualitative research methods, actively gathering primary-source information from battlefields, interviewing combat participants, and conducting desk research. We build results into data and evidence, write battle studies and other military analytic reports, brief, lead workshops, and run exercises.
WAR RESEARCHED BY HUMANS
Battle Research Group’s Artificial Intelligence (AI) Policy EXECUTIVE SUMMARY
n AI writing and editing reduce expertise, research value and trust in written work.
Both war and the research of war are inherently human endeavors. Understanding complex and uncertain information from battlefields requires a hands-on approach: We learn from the people who fight, who make key decisions and who are caught up in the fighting. Our field-research team members walk battlefields and see war from the perspective of the combatants. Our desk researchers immerse themselves in the complexity of offline and online information, applying their expert judgment to find and assess sources and to build complex narratives. Engaging in hard, detailed and sometimes mundane research tasks sharpens our expertise, sustaining and improving our value as researchers over time.
n AI-driven or -aided battle research will further reduce the fidelity and accuracy of war information over time, undercutting both AI efficacy and human research.
Artificial intelligence tools and particularly large-language models (LLMs) are ubiquitous. They offer many – and many rapidly improving – ways to enhance analysis and cut research time. Some artificial tools might help reduce known human weaknesses like bias and organic knowledge limitations, and they might significantly reduce research costs. However, as of early 2026 no valid longitudinal studies exist showing that AI use improves qualitative research and-or invalidates necessarily slower, more methodical approaches. While we see some utility for narrow technical tasks, we believe AI tools like LLMs are unsuitable for the core qualitative work of battle analysis and publication:
Moreover, research is our craft. Practicing our craft without artificiality gives us professional satisfaction which in turn motivates us to continue our present research and also our research careers. We enjoy making even small contributions to greater knowledge and, ideally, to at least minor improvements in policy. And we believe our expertise and the expertise of our many thousands of research colleagues around the world is both inherently valuable and relevant to human progress. Satisfaction from hard work well done reciprocally builds and sustains the human motivation to work hard and build expertise. Therefore, we will apply long practiced and broadly accepted human research methods. We will learn from our immersion in the complex and inconveniently messy details of war and, in doing so, continue to build our organic knowledge. By choosing this approach we hope to make small but relevant and reliable contributions to the preservation and growth of hard-won human expertise over time.
n AI cannot effectively judge relative value of inherently uncertain war information.
While we certainly will use electronic devices and may make some limited use of narrow AI tools for simulation and other activities – see explanation below – this is generally a long-term bet on human cognition and interpersonal engagement. Based on our collective years of research expertise, we believe it is both a necessary and proper bet both for our team and for our chosen profession.
n AI models add buried biases and inaccuracies that may be impossible to trace.
Given the rapid evolution of AI and related technologies the BRG senior staff will revisit this policy on a recurring biannual basis.
n AI research shortcuts can stunt the development of vital human expertise.
Battle Research Group Foundation, Inc.
March 23, 2026
1 of 19