AI is poised to transform the military strategy and defense industry of the United States, a major military power. The global military AI market is projected to grow at an average annual rate of 13% from 2025 to 2030. The US is developing AI-equipped unmanned surface and underwater vessels, as well as one-way long-range attack drones. It’s even said that with continued investment from the US and its allies, AI-powered weapons could be deployed within the next two years. However, the US isn’t the only country advancing the military use of AI. China is developing a wide range of unmanned systems. The China Ordnance Industry Corporation has unveiled a military vehicle capable of autonomously performing combat support missions at speeds of 50 km/h. In particular, China has consistently focused on developing (swarming) cooperative drones. The US also seems to perceive these drones as a threat.
The development of cooperative drones also appears to involve the use of insects. There is research into sending electrical signals to the nervous system of insects to freely control their movements. Cockroaches, in particular, are insects whose nervous systems are easy to modify and which are readily available. The technology involves attaching computer chips to the cockroach’s body, allowing it to be controlled like a remote-controlled toy. If it can fly using its unique insect wings, it can be treated much like a drone. On the ground, it can enter narrow pipes and gaps in rooms, enabling various investigations. The actions of these insect drones (robots) are not limited to a single individual. Many cockroach cyborgs can be released in large numbers at once. Using a group of drones would make them a useful tool in trench warfare. Aerial drones could be an effective weapon in trench warfare where tracking is difficult.
Scenario analysis for operational planning, which would take a human at least two days, can be completed in under a minute. The problem is that AI models tend to escalate suddenly compared to combat simulations conducted solely by humans. AI models have shown a tendency for wars, including nuclear warfare, to escalate abruptly. Using AI as a tool for waging war could even endanger the survival of humanity. The Go AI that defeated the strongest human Go player doesn’t explain why it made that move. Of course, humans aren’t staying silent either. Research is continuing to enable humans to interpret the reasoning behind the AI’s decisions. In reality, AI makes mistakes. When an AI makes a mistake, researchers are exploring what it was focusing on through trial and error. It seems that research is progressing by making it possible to visualize the computer’s circuitry’s reactions when it makes a mistake. There also seems to be a method of giving the AI many test cases that are likely to cause mistakes and observing how the AI behaves. We hope to improve these methods and pool our wisdom to stop AI from running amok.
