AI in warfare being tested in Iran, needs ‘much more’ careful thinking by NZ – Defence

0
6

Source: Radio New Zealand

Defence says new AI-supercharged weapon systems will need to be “very, very carefully designed”. NZDF / Supplied

New AI-supercharged weapon systems will need to be “very, very carefully designed” to comply with international and domestic laws, MPs have been told.

And it would be the software behind the systems that would dictate like never before just how effective any new missiles, guns or electromagnetic jammers were, a defence official told a select committee.

Defence ministry deputy secretary Anton Youngman said it was time for some serious thinking by New Zealand.

“One of the key points that we talk about here is that with these new capabilities … they need to be very, very carefully designed to comply with international and domestic laws,” he said.

The briefing coincided with the first week of the Iran war.

Experts said the war was testing out for real the questions of what artificial intelligence should be used in warfare and who controlled it.

Fox News has reported that the advance in AI “is changing the nature of the battlefield by speeding up targeting and analysing intelligence all while raising new concerns over the role of human judgment and oversight in modern warfare”.

The Guardian reported, “The use of AI tools to enable attacks on Iran heralds a new era of bombing quicker than ‘the speed of thought’ experts have said, amid fears human-decision-makers could be sidelined.”

AI targeting has been developing rapidly in the last several years.

Youngman, for his long-term insights briefing of the select committee, drew on a less militaristic example. He described a future where NZ kept an eye on nearby oceans by using satellites, drones flying high and on and under the sea, surveillance aircraft and land-based radars – ” all of these working in sync together”.

The software did that syncing.

Such technology was typically ‘dual-use’ with civilian and military applications.

Youngman went on: “The ability of defence forces to collect and analyse data at speed will increasingly be the key determinant of military advantage.”

Defence Minister Judith Collins in her speech to a geopolitics conference on Tuesday said New Zealanders understood the world had changed, and “the highly skilled personnel” in defence needed to be ready to do what the govenment “and people ask of it”.

“That’s why we are focusing on more than doubling our defence spend and investing in a defence force that is combat capable with enhanced lethality and deterrence; a force multiplier with Australia and increasingly interoperable with partners,” her speech notes said.

Defence Minister Judith Collins. Nick Monro

What does this have to do with NZ?

NZ has already put development of these syncing technologies on a faster track under last year’s $12 billion defence capability plan (though officials had been tightlipped about the aim to get a sovereign satellite).

Its latest move was to start testing 14 drones for the sea and air, with potential strike capability, from local firm Syos.

It was also working internationally through its defence science technology section with its counterpart in Australia, and with the US and other countries. NZ has not waited to join AUKUS Pillar Two – which focuses on emerging military tech – to make these moves.

AI-targeting experiments were part of that. The NZDF has been taking part in the US-led Project Convergence exercise to test joint AI systems alongside multinational forces.

Last year’s exercise in California had a “digital backbone” provided by data-mining firm Palantir.

The Washington Post has reported that Palantir tech was being used by the Pentagon in Iran. The Post said its targeting system called Maven was using an AI tool, Claude.

“Anthropic’s AI tool Claude central to US campaign in Iran, amid a bitter feud,” ran the paper’s headline.

Palantir, co-founded by Peter Thiel, a NZ citizen, has said the software used at Project Convergence “provided a unified data infrastructure for advanced battlespace management that empowered users across all levels to plan, execute, and assess operations effectively and enable commanders to rapidly make informed decisions”.

Palantir co-founder Peter Thiel. Marco Bello / Getty Images / AFP

Another such Convergence exercise was scheduled for the coming US summer. The NZDF did not respond when asked how many people it was sending.

RNZ has previously reported how this work fits under a Pentagon top-priority project with allies and partners called CJADC2 or Combined Joint All-Domain Command and Control.

‘Needs to be thought through much more carefully now than it ever previously did’

Youngman offered MPs another insight, that the ascendancy of software would change soldiering itself.

“Under the human-machine team … it’s a different role for defence personnel in this long-term future,” he said in response to National MP Tim Costley suggesting that NZ might be too small to properly deploy AI weapons and be better off adding to its soldiers, sailors and bullets.

Youngman said the role was moving potentially “from less kind of in the field work and more into that kind of tuning and training systems, interpreting the outputs, making decisions and ensuring adherence with … law and doctrine”.

Whose law and doctrine? That second question, of who controllrd the AI, also came up at the committee.

Green MP Teanau Tuiono asked, “You were saying earlier around making sure that the system design adheres to domestic international law. How are you going to do that?”

Green MP Teanau Tuiono. RNZ / Samuel Rillstone

Youngman said the challenge was new, now that machines could now take action themselves, for the first time in warfare.

“It’s going to continue to be a growing challenge and something that through the design of the capabilities, needs to be thought through much more carefully now than it ever previously did,” he said.

Labour MP and former Defence Minister Peeni Henare asked, “Do we have the foundational legislation to make sure that we’re able to govern effectively in the spaces of war?”

Youngman replied that was beyond the scope of the long-term briefing but added, “This is exactly the type of questions that this research is pointing to and saying we need to have this conversation.

“We are looking at a longer term horizon here, 2035, and the reason for doing this type of long-term research now is to say these are the types of conversations that we need to have.”

What about NZ being able to afford its own cloud-based AI military systems in future, Henare asked.

“Really good quesiton,” said Youngman. “I think the importance of remaining interoperable with partners is going to be key. It is today and it will continue to be.

“However … [the briefing] does talk about needing to continually balance that cost with sovereignty, with legality and social licence.”

‘A grey ship is a grey ship’

Everyone agreed that explaining all this to the public was much harder than talking about buying a new frigate.

“A grey ship is a grey ship,” said Henare.

“People will read this and go, this is preparing us for AUKUS,” he added.

Labour MP and former Defence Minister Peeni Henare. VNP / Phil Smith

Youngman replied that defence would “need to be more proactive” in communicating around the new capabilities.

The briefing itself said this was one of “three major shifts” defence had to get its head around.

“Public trust in defence forces is earned, not assumed. Ensuring Defence maintains public trust will remain essential, and possibly more challenging, in an environment defined by increased contestation and technological change,” it said.

When RNZ asked NZDF to lay out the nature of its technology and data-sharing with the US and other Five Eyes partners, Defence responded by turning it into an Official Information Act (OIA) request that would take at least five weeks to answer; similarly, a question about whether defence was taking a role in testing or developing systems from Palantir.

“Your request is noted, but the NZDF still needs to manage information requests in the way it deems appropriate,” Defence said.

The nature of NZ’s national security work within Five Eyes had come up earlier at a select committee. In that case, the SIS and GCSB replied they had tight controls around intellligence sharing and could withhold intel if legal, policy and human rights settings were not met.

An OIA in December showed that defence currently used nine AI-enabled tools in a restricted cacpacity for research in data and sensor processing and modelling. Sensors could be used in targeting.

The nine were: ChatGPT, Dalle-2, Github Copilot, Azure Machine Learning, Azure OpenAI services, Microsoft Copilot, Microsoft Teams, AiZynthFinder and Meta Llama 2.

National MP Dana Kirkpatrick thanked Youngman for the insights briefing: “There’s no time like the present in the current geopolitical challenges to be talking about future capability and interoperability in defence.”

Sign up for Ngā Pitopito Kōrero, a daily newsletter curated by our editors and delivered straight to your inbox every weekday.

– Published by EveningReport.nz and AsiaPacificReport.nz, see: MIL OSI in partnership with Radio New Zealand

Previous articleCricket: Fomer White Ferns captain Sophie Devine goes for top-dollar in The Hundred auction
Next articleNZ will release six days of fuel amid global concerns over supply