Abstract
An introduction to this issues’s special section on generative artificial intelligence, and its implications for defence and security.■
When generative AI came into the limelight with the release of ChatGPT in November 2022, the world’s attention suddenly focused on the implications of what looked like a technology so revolutionary that it could question not only what humans would be able to do with it in the coming years, but even whether its development would ultimately upend the economic, social and political foundations on which our current domestic and international orders are based. Much has been said and written and opinions shared (some with more of an evidence basis than others) in the intervening time, yet the jury is out on whether AI–and specifically large language models–are truly revolutionary, or simply an added tool whose efficacy will ultimately depend on the humans who use it and how they choose to deploy it. While the debate continues, however, policymakers and practitioners still need to make decisions–about whether to regulate the development and use of these technologies, or whether, and how, to procure them and integrate them in existing systems or thinking. The question becomes even more pressing when grappling with systems that are indeed able to deliver impressive results, without the experts who devised or used them being able to understand how exactly they can do what they do in the first place. There is therefore an underlying question about whether these systems should be used at all, if their human users–even the most technically proficient ones–cannot detail the process by which they came to be able to do what they do. Other issues are also pressing: practically, what this technology can deliver in the immediate to medium term, and how we should prepare and adjust existing processes and approaches to accommodate them.
Without detracting from the fact that some types of AI have been used in multiple defence functions for several years, the question of how to deploy generative AI is no less pressing for armed forces, intelligence agencies and security actors the world over. We asked experts in some of the key fields of defence and intelligence to reflect on this question, and explore some of the ways in which they can see generative AI contributing to existing and future activities. The contributions in this section are therefore initial reflections–sometimes sparked by interactions with generative AI tools themselves–on what is sure to be a growing debate on the very practical applications of large language models in key security fields. Rather than definitive answers, they are meant to focus the mind and provoke reflection–and perhaps suggest some further avenues for research and practical experimentation.■
The growth of generative AI? Generated by AI. Courtesy of Adobe Firefly
Additional information
Notes on contributors
Emma De Angelis
Emma De Angelis is the Editor of the RUSI Journal.