For some, the idea of the military using Generative AI(GenAI) might conjure up a mental image of a commander asking ChatGPT to generate ten alternative battle plans, or asking it to identify the top five high-value enemy targets for the next five sorties. Whilst simplistic, these scenarios are not so far away from the truth.
In its favour are Generative AI’s speed, ease of use and ability to draw on vast volumes of data to generate responses to queries. On the other hand, the very public lessons that ChatGPT has taught us about GenAI’s limitations include ‘hallucinations’, its preference for historical versus real-time data, its unreliable mathematics functions, and risks to data security.
However, large language models (or LLMs) are simply too big for the military to ignore. They can be astonishingly fast to generate useful responses to complex queries, can do so in natural language (of multiple languages, in fact), and have the ability to present and compare seemingly endless sets of data.
So, it is an area of artificial intelligence that warrants urgent examination and experimentation by the military. The U.S. Department of Defense formed Task Force Lima over the summer, to play the lead role in analysing and integrating generative AI tools across the DoD. According to the DoD, GenAI has significant potential to improve intelligence, operational planning, and administrative and business processes. It obviously also emphasises careful implementation and a focus on security, in order to manage associated risks effectively.
It’s difficult to narrow down where generative AI might be of most use to the military today and that’s for two main reasons. Firstly, because the technology will have use cases everywhere, from catering supplies inventory, through to battlefield decision support. Secondly, this is a fast-changing emerging technology with capabilities that are growing every day. Just because a aspect of functionality is missing or less reliable today, doesn’t mean that it won’t be created or enhanced tomorrow.
In the short term, we’re likely to see generative AI used to improve efficiency and productivity, and reduce costs in administration and business affairs, internal communications, logistics and supply chain, human resources, training, healthcare and cybersecurity. However, in the near future, we’re sure to see GenAI applications created for operational planning, intelligence analysis and battlefield decision support.
The stakes are too high not to find safe, reliable ways of harnessing GenAI. For example, one day soon, a commander will be able to contrast and compare battlefield scenarios based on historical data, current data and predictive analysis in just a few seconds. An adversary’s movements, offensives and targets could be analysed, it’s communications translated and annotated, and yes, the potential value and recommended priority of military targets could be provided: all in just a few seconds. These are not capabilities that you stand by and leave your adversaries to capitalise on.
by Carrington Malin
