This week, officials from more than 100 countries and territories participated in the International Conference on Autonomous Weapons in Vienna, Austria, to discuss the integration of artificial intelligence and military technology. The Associated Press commented that a "third military revolution" after gunpowder and nuclear weapons is likely to occur as AI-powered lethal autonomous weapon systems (LAWS) are put into practice.
Austrian Foreign Minister Schallenberg warned that "the decision of who lives and who dies must be in the hands of humans, not machines." But the truth is that technology is advancing at a rapid pace, and policies are lagging behind. Humanity is facing an Oppenheimer moment. ”
The Israeli army tests an AI combat system in the field in Gaza
From April 29 to May 1, the International Conference on Autonomous Weapons was held in Vienna, Austria, under the theme "Humanity at the Crossroads".
Austrian Foreign Minister Schallenberg: Humanity is at a crossroads, and autonomous weapon systems will soon fill the world's battlefields. We've seen this through AI (controlled) drones and AI-based target screening.
Al Jazeera: An investigation revealed that the Israeli Army has been using untested AI tools during the bombing campaign in Gaza.
On April 3, according to the Israeli media "+972" magazine, the Israeli military is testing an AI combat system called "Lavender" in Gaza. The system can quickly determine the when, where and how to kill the target.
According to Al Jazeera, up to 37,000 Palestinians have been targeted by the Lavender system. According to an insider in the Israeli Army, the "Lavender" system is currently rapidly learning and improving itself in actual combat. The ideal goal of the Israeli army is to limit the number of civilians killed incidentally to 100 in the event of a successful hunt for a senior Hamas official.
Yuval Abraham, an Israeli investigative journalist who disclosed the incident, told the media that the "lavender" system played a central role in the current round of bombing of Palestinians.
Yuval Abraham, Israeli investigative journalist: After October 7, 2023, the Israeli military made a decision that tens of thousands of people could now be bombed inside their homes. This means that not only will they be killed, but everyone in the building, including children, family members. In order to achieve its goals, the Israeli army relies on this AI machine called "Lavender", but there is very little human supervision of it. Insiders have admitted that he is a rubber stamp relative to the decisions made about the machine.
CNN reported in December that nearly half of the munitions Israel dropped on Gaza in the first two months of the conflict were unguided bombs that lacked precision. Bombs often destroy entire buildings, causing enormous damage to civilians. Israeli investigative journalist Abraham said that according to AI's analysis, some Hamas targets are "not important enough" and that in order to save costs, there is no need to waste expensive guided missiles to precisely hit the specific floor where the target is located. In the early stages of the war, this caused a large number of innocent casualties.
Yuval Abraham, Israeli investigative journalist: I mean, under the terms of international law, when you design a system that targets 37,000 people, and you verify that 10 percent of them are not actually militants, who are not closely linked to Hamas or have nothing to do with it, but you still authorize the system to launch for weeks without oversight. I mean, isn't that a violation of the principles [of international law]? I think international law is facing a crisis right now, and AI systems are making it worse.
According to Anthony, the author of the book "Palestine Proving Ground" and a reporter for the Sydney Morning Herald, Israel's bold, illegal and unethical training of AI weapon systems is likely to be supported by a large amount of data and intelligence.
Anthony, reporter for the Sydney Morning Herald: I suspect the source of intelligence could be the US or Australia. Pine Gap is a major U.S. military base in central Australia, through which the U.S. provides information to Israel. All of this data is fed into the ("lavender") system, resulting in a large number of targets. Israel is currently trying to sell these (AI) tools to foreign entities and governments. These Governments have not questioned Israel's actions in Gaza, but have applauded them.
The battlefield in Ukraine has become an "AI weapons laboratory" for the US military
In the eyes of Lieutenant General Donahue of the 18th Airborne Corps of the United States, who was in charge of AI combat tests on the battlefield in Ukraine, the battlefield in Ukraine has become the "AI weapons laboratory" of the US military.
On April 23, the latest progress of the Pentagon-led "Mayvin Plan" was revealed.
According to the New York Times, the Meiwen intelligent system is being rapidly trained and perfected on the battlefield in Ukraine. To put it simply, the AI system can aggregate images from satellites, radar, infrared devices, and even social media into a single screen, helping people connected to the system quickly identify targets. By plugging into a weapon system such as a fighter jet or a drone, the human commander only needs to make the final decision to confirm the strike.
Mike Martin, former British Army officer: One of the manifestations of AI in weapon systems is autonomy. The reason we want weapons systems to be autonomous is because we want them (weapons) to strike quickly. In war, if you can strike first, then you have a huge advantage. Intelligence analysis is an effective way to identify (hit targets), and identifying a certain target is exactly what AI systems are very good at.
For example, if you ask the AI system to look at a photo of a tank, the system will locate its location and the drone can strike at the tank.
According to the British "Daily Mail", at least 10,000 drones are flying over the Russian-Ukrainian front line at any one time, and most of these drones cost no more than $2,000 each, but they can destroy a multimillion-dollar main battle tank.
In October last year, according to Forbes News, the Ukrainian drone company Sakr claimed that it had deployed a fully autonomous drone, the Saker Scout. Powered by AI, it is able to decide for itself who to kill on the battlefield. Thacker officials claimed that the drone had carried out small-scale autonomous attacks.
Intelligent systems accelerate the pace at which machines become decision-makers on the battlefield
According to the British Channel 4 News Report, the platform of the Meiwen intelligent system was built by the Pentagon's contractor, the American company Palantaire Technologies. The tech company openly claims that its AI system is the digital brain of Ukrainian weapons, and that most of the targets of the Ukrainian army are locked and hit with the help of this platform.
At the end of last year, Palantaire released details of the latest generation of AI platforms. The platform introduces large language models and cutting-edge AI technology for military design. According to Palantaire, the tool is like a chatbot in the military field, providing commanders with real-time strategic decision-making suggestions through simple voice command input. Critical military decisions that could have taken hours now take minutes.
Human Operator: Generate three action plans to target the enemy equipment, generate a route from a small detachment Ω to the enemy equipment, and destroy the enemy equipment.
Although Palantaire says that their systems follow ethical and legal guidelines, are run with human oversight, and cannot be attacked independently. However, some analysts have pointed out that the technology to create fully autonomous weapons has long existed, and speed is a decisive factor on the battlefield, and platforms like the Meiwen system have undoubtedly accelerated the pace of "making machines decision-makers on the battlefield".
Paul Schaar, a former U.S. military officer, was named one of Time magazine's 100 most influential people in the field of artificial intelligence. In February, he warned in Foreign Affairs magazine: "Without strict control of autonomous weapon systems, the role of humanity in the future will be reduced to turning on machines and sitting on the sidelines, with little power to control or even end wars." ”
According to Bloomberg, in the air raids and bombings carried out by the US military in the Middle East in early 2024, the Meiwen intelligent system helped the US military lock on at least 85 targets, covering Iraq, Syria, Yemen and the Red Sea region.
The U.S. military is accelerating the advancement of artificial intelligence technology
According to Time magazine, the U.S. military's investment in artificial intelligence increased dramatically in fiscal year 2024.
According to the U.S. government budget for fiscal year 2024, including the Department of Defense, the Department of Energy, the Department of Homeland Security, and other agencies, a total of more than $251.1 billion has been invested in the AI field. If the government's external financing and capital market investment are included, the U.S. investment in AI is expected to exceed trillions of dollars in 2024.
In November 2023, the U.S. Department of Defense released the "Artificial Intelligence Implementation Strategy". The summary points out that the U.S. Department of Defense is accelerating the implementation of advanced artificial intelligence technology to ensure that the U.S. military maintains a decision-making advantage on the battlefield in the coming years.
On May 2, at Edwards Air Force Base in California, U.S. Air Force Secretary Kendall personally experienced an artificial intelligence warplane. He flew in an X-62A VISTA fighter jet piloted by AI.
US Secretary of the Air Force Kendall: The computer doesn't get tired, it's not afraid. It will follow its rules and move as close to perfection as possible, while there will always be some variables in human performance.
According to the U.S. Defense News Network, on April 20, the Defense Advanced Research Projects Agency (DARPA) released a video of a simulated air battle between an AI-piloted VISTA fighter and an F-16 fighter piloted by a senior pilot. This is also the first air battle between AI and humans.
In the face of the media, US Air Force Secretary Kendall refused to disclose the specific situation of this air battle. But he said that more scenarios will be held in the future, so that AI fighters can polish their technology.
U.S. SECRETARY OF THE AIR FORCE Kendall: We can make it [the AI fighter] the vanguard, we can put it in some deliberate place to sacrifice in order to attract firepower or find out where the enemy is. Our pilots generally believe that this will give them a competitive advantage. We're taking advantage of (AI) technology. It's one of the keys to this game, and the stakes are high.
Palmer Larkey: "Oppenheimer of Silicon Valley"
In this regard, the "Russia Today" television station bluntly said that the war is creating a large amount of "wealth" for the US military industry.
As conflicts and confrontations intensify in many parts of the world, the demand for AI-powered weapons is increasing, and whether it is Loma, RTX (formerly Raytheon) or General Dynamics, whether it is industry giants or start-ups, are stepping up their autonomous weapons development projects in hopes of getting a piece of the huge defense budget in the United States.
Palmer Rakhi, Founder and CEO of Anduril Industries: This is "rage". It is an unmanned fighter that is capable of working with manned fighters and other manned aircraft, greatly expanding the range of targets that can be attacked without putting personnel at risk. This is one of the coolest things about autonomous weapon systems.
Palmer Lachi, 31, is the founder and CEO of Anduril Industries, a U.S. defense technology and weapons manufacturer, and is known as the "Oppenheimer of Silicon Valley" by the U.S. media.
On April 24, Anduril was selected by the U.S. Air Force to build and test a drone prototype for the next phase of the AI Cooperative Combat Aircraft program. The Wall Street Journal notes that against the backdrop of a near-monopoly of the industry by the five largest US arms manufacturers, Anduriel is the only defense start-up that has won multiple contracts from the Pentagon.
In 2012, at the age of 20, Rakki founded Oculus VR, a virtual reality company, and two years later, he sold it to Meta for $2 billion. In 2017, at the age of 25, Rakhi founded Anduril to devote himself to the development of AI-powered weapons, and a few months later signed a major defense order with the Trump administration to provide AI surveillance towers for the Mexican border.
Today, its autonomous weapons systems have been procured by multiple departments of the U.S. military. So far, thousands of Altius-600M attack drones have been supplied to the battlefields in Gaza and Ukraine. At the end of last year, the company was valued at about $8.48 billion, which is only about one-tenth of that of traditional American military giants, but the Wall Street Journal concluded that Anduril's recipe for success is low cost, easy operation, and high product development efficiency.
At the end of last year, Anduril launched the Roadrunner high-explosive drone interceptor, which is only 1.5 meters high, flexible and has a warhead load three times that of similar aircraft. Compared with the AGM-114 Hellfire missile, which can only be used once and costs $150,000, and the AIM-120 missile, which costs only $180,000, the "Walking Juan", which costs only tens of thousands of dollars, can take off and land vertically from the launch platform. If it turns out that the threat is non-existent, it can be returned to the designated location in good condition.
What's even more impressive is that, unlike the complex hardware platforms of traditional military giants, Anduril's operating platform, Lattice, acts as a command center, allowing a single operator to control and coordinate multiple AI autonomous weapons, whether it's drones, submarines, or mobile surveillance towers.
Chelan, Head of Platform R&D, Anduril Industries: You'll see all the assets of this team at the bottom, and then you'll see a graphical display. It looks very similar to the software we use in our daily lives. There's a map in the background, and you'll see multiple devices on the app, and you'll be able to see the role each one plays. Engineers design our software just like software on a mobile phone or iPad, and it's very user-friendly to use.
Palmer Rakki has often said that Tesla's self-driving is better than U.S. military vehicles, and that the iPhone's computing power is more powerful than systems commonly used by the U.S. Department of Defense. The operation of the AI weapon system of the future should be as simple as operating any smartphone. According to Rakhi's philosophy, on the battlefield, machines can do better, so why labor people?
Palmer Rakhi, Founder & CEO, Anduril Industries: You can't rely on humans to do everything when you need to micromanage thousands or tens of thousands of weapons systems running simultaneously in an adversarial environment. Some task bots are better at it.
The Wall Street Journal bluntly said that Palmer Lachi's business model is expected to reshape the U.S. defense industry. However, Rakhi's philosophy is contrary to many researchers.
This is reminiscent of the "Killing Robots" video made by Stuart Russell, a professor in the Department of Computer Science at the University of California, Berkeley, in 2017. The video is considered by the industry to predict the future of AI weapons.
The video shows how in the near future, micro-drones equipped with facial recognition systems will be equipped with explosives and become terrifying "AI killers". Drones have the ability to think autonomously, and they first attach themselves to buildings, blasting holes in walls so that other drones can enter and hunt down specific students. Drone strikes at more than a dozen universities have killed thousands of university students. And the same tragedy occurred in the US Congress, where the senators chosen by the AI were killed.
Russell, Computer Scientist, University of California, Berkeley: Allowing machines to selectively kill humans would be devastating to our safety and freedom. Thousands of my peers think we have a chance to stop the future you've just seen, but the window of time to act is closing fast.
On May 1, at the International Conference on Autonomous Weapons, Aguirre, a well-known American information physicist, warned that the "era of killer robots" predicted by Dr. Russell has arrived. ”
Now, with a 3-D printer and basic programming knowledge, a student can build a drone capable of causing mass casualties. Autonomous weapon systems have changed the concept of international stability forever. Hacking of AI systems, loss of satellites and other data, and even hallucinations caused by AI itself can be potential risks for AI weapon systems.
- Anthony Aguirre, information physicist
Austrian Foreign Minister Schallenberg: Actually, in a way, it's worse than the Oppenheimer moment. This is the biggest revolution on the battlefield since the invention of gunpowder, and this technology is cheaper and develops faster.
Christopher Nolan, the director of the movie "Oppenheimer", once said that many top AI researchers regard the present as their "Oppenheimer moment". It is an ominous premonition that the experience of the past is permanently invalid, and that an unknown catastrophe is about to befall in an instant. At this moment, the fate of mankind and the earth seems to be retracted into a blind box.
This week, Alexander Kment, Austria's high-level official for disarmament affairs and the mastermind of the autonomous weapons conference, pointed out that it will be difficult to rely on diplomatic means to solve the regulation of AI weapons in the short term, because it will be a tome treaty.
How to usher in the "Oppenheimer moment" in the era of artificial intelligence requires global wisdom.
Source: CCTV News