How generative AI can improve military training

Generative artificial intelligence (AI) has been proving itself as an efficient tool in various domains. In the defence realm, it has the potential to improve military readiness and training, experts believe. 

By using models that can recognise patterns in massive datasets and use those patterns to generate statistically probable data, armies could better analyse adversaries' behaviour and predict their strategies as well as the potential ways they would attack and defend.

Those models can also provide realistic simulation environments for skill-building practice while reducing the costs and risks associated with large-scale training events.

Related Articles

US to increase funds for artificial intelligence capabilities in FY2024

I/ITSEC 2023: How VR and AR are reshaping the training capabilities for the Pentagon

I/ITSEC 2023: US Marine Corps seek AI capabilities for training purposes

Another advantage of generative AI is the possibility to scale and accelerate the deployment of simulation systems as well as quickly update capabilities and better prepare troops for deployment in diverse challenging scenarios.

Speaking at the I/ITSEC 2023 event in November, Keith Brawner, programme manager at the Institute for Creative Technologies, a DoD University Affiliated Research Center (UARC) sponsored by the US Army, explained that generative AI ‘will be critical’ for training and simulation.

‘We were already using generative trained models in order to do adversarial actions within simulations as well as using them to generate a three-dimensional terrain,’ Brawner noted.

Andy Van Schaack, associate professor of the Practice of Engineering Management at Vanderbilt University, pointed out that ‘we are on the threshold of the most significant technological development in the history of civilisation’.

‘Intelligence is power,’ Van Schaack claimed. ‘The only thing more impressive to me than the current capabilities of generative AI are the rate of [their] progress.’

Svitlana Volkova, chief AI scientist at Aptima, which is a supplier of AI solutions, highlighted that in the short-term military training should use different agents to support learners, instructors and leaders.

‘In five years, I predict more integration of automation into the workflows,’ she claimed.

Paratroopers conduct simulated missions at Fort Bragg. (Photo: US Army)

Aiming to harness the potential of this type of technology and speed up its deployment, in August the DoD established a generative AI task force to enable its use alongside associated large language models (LLMs) in a responsible and strategic manner by minimising risks and ensuring national security.

Named Task Force Lima, it is intended to enable overcoming issues related to a broader use of generative AI across the department.

However, enhancing their use for training purposes might require technological advances, such as faster microprocessors, additional memory and better algorithms and training datasets.

From Volkova’s perspective, it is also ‘very important’ to focus on technologies that enable the development of ‘safe, secure and trustworthy artificial intelligence systems’.

Van Schaack highlighted that China represents a risk for the US in this realm, as Beijing has ‘much work being done in the field of AI’.

‘We would be naïve to think that adversaries of the United States are not capable of beating us in many ways.’ Van Schaack remarked. ‘They have very, very smart people that are very eager and focused,' he noted.

The DoD also conducts other acquisition and development programmes involving AI. This is the case with the Data Readiness for Artificial Intelligence Development (DRAID) initiative, which seeks to cover the entire data preparation lifecycle leading to system creation and model training.

Tradewind is another Pentagon effort. Designed to enable the department to access small business innovation. This initiative has provided the DoD with over 300 data and data integration solutions.

Pentagon agencies have been increasing efforts to enhance the use of AI tools. (Photo: US Army)

The DoD also works on the TryAI programme, which is intended to enable the Pentagon to quickly test industry solutions before initiating a purchasing process.

Additionally, branches and agencies have been increasing efforts to enhance the use of AI tools. The US Army, for instance, has been accelerating the procurement and integration processes of AI and machine learning (ML).

Army Futures Command’s AI Integration Center (AI2C) has been part of the overall effort, working on building up relations with academic, industry and professional partners for AI-related education and R&D.

The service has also been conducting several experiments, tests and demonstrations involving AI and ML use under Project Convergence.

Another army line of action includes partnering with academia to develop advanced systems. In this space, the branch has been working alongside the USC Institute for Creative Technologies (ICT) to develop the Game-If-AI app.

The solution is intended to teach soldiers in technical and non-technical areas how to use AI tools.

The USN, meanwhile, has its Center for Applied Research in Artificial Intelligence (NCARAI), which works on basic and applied research on AI, focusing on intelligent agents, human-machine teaming and autonomous systems among others.

From the USAF’s side, the Academy Center for Cyberspace Research (ACCR) conducts initiatives on AI applied to swarm area defence and ISR.

DARPA also has projects in this domain which are centred on developing next-generation capabilities.