NATO: Future conflicts may be won or lost by AI
“We are really convinced that ongoing and future conflicts may be won, lost or heavily impacted by AI speed, AI efficacy and who is actually using AI in the battlefield.”
So said NATO’s head of data and artificial intelligence (AI) policy Dr Nikos Loutas at the AI Summit London during London Tech Week last month.
Loutas revealed that there were several examples in the ongoing crisis in Ukraine “where we see data and AI being used by both sides – but primarily by Ukraine”.
“Data and artificial intelligence are areas of strategic competition at the global level,” Loutas explained. “They are a political priority, and they concern each and every one of us.”
According to Loutas, the main challenge is the need to manage huge volumes of data and he asserted that, for NATO, data is a strategic asset to be used to maintain technology superiority.
Driven by leadership and the NATO alliance, ministers from 30 different sovereign states came together last October and agreed two flagship policy initiatives: the NATO Data Exploitation Framework Policy and the NATO AI Strategy.
Loutas said that these two complementary policy initiatives outline how data exploitation and AI can be applied to defence and security in a protected and responsible way, relying on standards of responsible use of data and AI, in accordance with international law and NATO’s values.
Implementation of both policies is currently underway, with the first products already being delivered.
“These initiatives are in tandem and are two sister initiatives that are aiming for the same objective: to help us better prepare for using data, doing more with data and becoming more mature with data and AI,” Loutas added.
The policies are leadership-driven initiatives, but they are also supported bottom up. “That’s important,” Loutas emphasised.
“In an alliance, it’s very important to bring together the practitioners across the organisation and make sure they are empowered to cross-pollinate other functions and parts of the organisation to help increase adoption.”
NATO’s data policy governs the use and management of data, whether NATO produced, acquired, open sourced or data shared by other nations. “The framework looks at the full spectrum of data use up to artificial intelligence. That’s why we think AI is different,” Loutas explained.
“There are political and practical reasons for carving out a dedicated strategy. There’s an opportunity for NATO allies to work together and accelerate efforts,” he added.
NATO’s AI strategy meanwhile attempts to accelerate and mainstream the use of AI via a series of use cases. “We’re convinced and determined to do AI for today and not just AI for tomorrow,” said Loutas.
“We’ve identified a collection of use cases at different levels of maturity. We’re working together with industry, allies, and partners to develop and further those use cases. Some are purely experimentation and others are about capability development.
What is important is that all those use cases address specific needs, operational priorities of the alliance, looking at everything from multi-domain operations to addressing climate change, to looking at the information environment and tapping into what’s being said there, cyber threat analysis, and enhancing our situational awareness,” said Loutas.
NATO’s AI policy also looks to identify and safeguard against the malicious use of AI by state or non-state actors.
Responsible Use of AI
According to Loutas, the UN’s use of AI needed to be “driven by our principles of responsible use”.
“That is a true commitment and something that we’re taking very seriously. Within our AI Strategy there are six principles introduced and agreed by 30 sovereign states.
“These principles are based on the widely accepted ethical, legal and policy commitments of NATO and complementary to national principles,” he said.
“They provide coherence and enable interoperability, as NATO and Allies use AI in the context of defence and security. The focus is now on putting them in practice.”
In terms of governance, Loutas asks: ‘What is the right governance for AI and putting the principles in practice?’ “We’re considering the creation of a specific board or committee that will look at how the principles are being implemented in the context of use cases. So, governance is one thing,” he said.
“We’re also looking a lot at standardisation, both process and technology standards, which can help us ensure and assure our applications are aligned and implement the principles,” he added.
“For that, we’re looking at working with standards organisations, the international community, and Allies and Partners. Such standards cannot be developed only within and for NATO. It’s important that our voice is heard and being considered.
NATO is also looking at test centres in the alliance, he shares. There’s one in the UK for example that will help NATO with validation and verification of AI applications.
“We’re also looking at technologies that can help us identify and treat specific risks related to those principles in the different models. Tools that can identify bias or can stress test models,” he added.
Loutas stresses that NATO cannot achieve all these objectives alone. “This is something that we must do together. We look at industry, academia, allies, and partners to work together,” he said.
“We’re not on an island or silo trying to figure things out by ourselves. It’s important to work with industry and equally important to work with the start-up ecosystem, which can see defence and security as a strange area,” he added.
A defence innovation accelerator network has been set up across the NATO Alliance, aiming to attract dual-use technology innovators and share its challenges with them and help them develop solutions for NATO.
Loutas revealed that there will be funding made available and there will also be possibilities for the “successful solutions to be opened to a market of up to 30 countries”.
“There’s a lot of interesting features for the start-up community and we can engage real experts that create tailored solutions for us,” he added.
The insights came a month after NATO launched a strategic initiative on Artificial Intelligence (AI) Horizon Scanning aimed to better understand AI and its potential military implications.
Subscribe to our Editor's weekly newsletter