AI in war being tested in Iran, New Zealand needs ‘further’ careful consideration – Defense

New weapons systems equipped with AI will need to be “very carefully designed” to comply with international and domestic laws, MPs said.

And, more than ever, it will be the software behind the systems that will determine how effective new missiles, guns and electromagnetic jammers are, defense officials told the task force.

Ministry of Defense Deputy Secretary Anton Youngman said it was time for New Zealand to think seriously.

“One of the important things we talk about here is that these new features need to be very carefully designed to comply with international and domestic laws,” he said.

The press conference coincided with the first week of the Iran war.

Experts said this war is really testing the question of what kind of artificial intelligence should be used in war and who should control it.

FOX News reported that advances in AI are “changing the nature of the battlefield by speeding up the targeting and analysis of information, while raising new concerns about the role of human judgment and oversight in modern warfare.”

The Guardian reported: “The use of AI tools to enable the Iran attack heralds a new era of bombing faster than the ‘speed of thought’ of experts, amid fears that human decision-makers will be sidelined.”

AI targeting has developed rapidly in recent years.

Mr. Youngman cited a less militaristic example in his explanation of the task force’s long-term insights. He described a future in which New Zealand monitors nearby oceans using satellites, drones flying over and under the sea, reconnaissance aircraft and land-based radar, “all working together”.

The software did that synchronization.

Such technologies were typically “dual-use” for civilian and military applications.

Youngman continued: “The Defense Force’s ability to rapidly collect and analyze data will be a key determinant of military advantage.”

Speaking at the Geopolitical Conference on Tuesday, Defense Minister Judith Collins said New Zealanders understood that the world had changed and that “highly skilled talent” in the defense sector needed to be ready to respond to “the demands of government and the public”.

“That’s why we are more than doubling defense spending and focusing on investing in a defense force that is combat capable with increased lethality and deterrence – doubling our military strength with Australia and increasing interoperability with our partners,” the speech notes said.

What does this have to do with New Zealand?

New Zealand is already accelerating the development of these synchronization technologies under last year’s $12 billion defense force plan (though officials have been tight-lipped about its goal of acquiring a sovereign satellite).

The latest move was for local company Syos to begin testing 14 sea and air drones with potential attack capabilities.

We also collaborated internationally with our Australian counterparts, the United States, and other countries through our Defense Science and Technology Division. NZ has not made these moves pending its participation in AUKUS Pillar 2, which focuses on emerging military technologies.

Experiments targeting AI were also part of this effort. The NZDF is participating in the US-led Project Convergence exercise to test joint AI systems with multinational forces.

An exercise in California last year used a “digital backbone” provided by data mining company Palantir.

washington post reported that Palantir’s technology was being used by Iran’s Department of Defense. The newspaper reported that the company’s targeting system, called Maven, used the AI ​​tool Claude.

The newspaper’s headline said, “Anthropic’s AI tool Claude is at the center of US operations in Iran, amid bitter feud.”

Palantir, co-founded by New Zealand national Peter Thiel, said the software used in Project Convergence “provided an integrated data infrastructure for advanced battlespace management, enabling users at all levels to effectively plan, execute and evaluate operations, and empowering commanders to make faster, more informed decisions.”

Another similar convergence exercise was scheduled for the upcoming U.S. summer. The NZDF did not respond to inquiries about the number of people dispatched.

RNZ previously reported how this effort fits into the Department of Defense’s top priority project with allies and partners called CJADC2 (Combined Joint All-Domain Command and Control).

“We need to consider this more carefully than ever before.”

Youngman offered another insight to lawmakers: Software superiority will change the soldiers themselves.

“Under human and machine teams…the role of the defense workforce in this long-term future is different,” he said in response to Tim Costley MP, who suggested New Zealand was too small to adequately field AI weapons and might be better off with more soldiers, sailors and bullets.

Youngman said the role “could move from field work to coordination and training systems that interpret results, make decisions, and ensure compliance with law and doctrine.”

Whose laws and doctrines? A second question, who controls the AI, was also raised by the committee.

Green MP Tenau Tuiono asked: “You said earlier that you would ensure that the system design complies with national and international law. How are you going to do that?”

Youngman said the challenge was new, as it was the first time in war that machines could act on their own.

“This continues to be a growing challenge and requires much more careful consideration through feature design,” he said.

Labor MP and former defense minister Pini Henare asked: “Are there fundamental laws to ensure that we can govern effectively in a theater of war?”

Youngman responded that that was beyond the scope of a lengthy briefing, but added: “This is exactly the kind of question that this study is pointing to and saying we need to have this conversation.”

“We’re thinking long-term here, to 2035, and the reason for doing this kind of long-term research is: now That means we need these kinds of conversations. ”

Mr Henare asked what would happen if New Zealand were able to purchase its own cloud-based AI military system in the future.

“That’s a really good question,” Youngman said. “I think it’s going to be important to maintain interoperability with our partners, that’s what’s going on, and that’s always going to be the case.

“but … [the briefing] It notes the need to continually balance costs with sovereignty, legality, and social license. ”

“A gray ship is a gray ship.”

Everyone agreed that explaining all this to the public is much more difficult than talking about the purchase of new frigates.

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

“People will read this and leave. This is preparation for AUKUS for us,” he added.

Youngman responded that defenders “need to be more proactive in communicating” about their new abilities.

The press conference itself said this was one of the “three big changes” that defenses had to worry about.

“Public trust in the Defense Forces is earned, not assumed. It will continue to be essential for the Defense Forces to maintain public trust, and may likely become more difficult in an environment defined by escalating conflict and technological innovation,” the report said.

When RNZ asked the NZDF to explain the nature of its technology and its data sharing with the US and other Five Eyes partners, Defense turned this into an Official Information Act (OIA) request and said it would take at least five weeks to respond. Similarly, questions regarding whether the defense sector played a role in the testing or development of Palantir’s systems.

“While your request has been noted, the NZDF must still manage information requests in the manner it sees fit,” the Ministry of Defense said.

The nature of NZ’s national security activities in the Five Eyes had previously been addressed by the select committee. In this case, SIS and GCSB responded that they have strict controls over information sharing and can withhold information if legal, policy and human rights settings are not met.

The December OIA indicated that the Department of Defense currently uses nine AI-enabled tools in a limited capacity for data and sensor processing and modeling research. Sensors may be used for targeting.

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

Congressman Dana Kirkpatrick thanked Mr. Youngman for his insight briefing and said, “With the current geopolitical challenges, there is no better time than now to have a conversation about future capabilities and interoperability in defense.”

Sign up for Ngā Pitopito Korero. Our daily newsletters, hand-picked by our editors, delivered straight to your inbox every weekday.