/View.info/ Humanity has entered the era of algorithmic wars. This was announced in December 2020 by Assistant Secretary of the Air Force Will Roper in an interview with Popular Mechanics, talking about the successful use of an artificial intelligence (AI) system as the co-pilot of the U-2 Dragon Lady aerial reconnaissance aircraft. According to an Air Force press release, an artificial intelligence algorithm called ARTUµ was responsible for sensory control and tactical navigation during this flight.
The pilot and ARTUµ conducted a reconnaissance flight during a simulated missile strike. The main task of the AI was to find enemy launchers, while the human pilot searched for enemy aircraft. According to Roper, ARTUµ is a modified version of a gaming algorithm that outperforms humans in games like chess and go. The flight results were so promising that the electronic warfare function was added to ARTUµ’s next mission.
„Like any pilot, ARTU has strengths and weaknesses,” writes Roper. “Understanding them to prepare both humans and AI for the new era of algorithmic warfare is our next imperative. We either become science fiction or we become history.”
„To fight and win in a future conflict with an equal adversary, we must have a decisive digital advantage.” Air Force Chief of Staff Gen. Charles Brown said. “Artificial intelligence will play a critical role in achieving this advantage.”
The development of combat algorithms in the United States began in April 2017 with the creation under the auspices of the DIA (Defense Intelligence Agency, DIA) of the Combat Algorithmic Warfare Cross-Functional Team, known as the Maven project.
The project is designed to improve the efficiency of air reconnaissance data analysis. The project’s first successful experiment was the analysis of video data from the MQ-1C Gray Eagle and MQ-9 Reaper UAVs. A reconnaissance drone delivers terabytes of intelligence daily. Before AI was included in the analysis of this data, the analytics team had to work 24 hours a day to use only a fraction of the sensor data from just one drone. AI, specifically ARTUµ, does this in seconds.
In 2018, the US Department of Defense published a report “Artificial Intelligence. Strategy. Using AI to promote our security and prosperity ”( Artificial intelligence. Strategy. Harnessing AI to promote our security and prosperity), which outlines the Pentagon’s views on using the potential of AI technology in algorithmic warfare.
The Joint Artificial Intelligence Center (JAIC), established in 2018, is called to implement artificial intelligence systems in all structural units of the US military. Jack Shanahan, former head of the Maven project and chief ideologue of the algorithmic wars, headed JAIC.
JAIC’s main area of interest is Neuromorphic computing. The term “neuromorphic” in reference to computing systems means that their architecture is based on the principles of the brain. The neuromorphic system is a departure from von Neumann’s classical computer architecture.
In the von Neumann architecture, computing blocks and memory blocks are separated. The main difference between the architecture of neuromorphic processors and the classical architecture is that they combine memory and computing cores, and the data transfer distance is minimized. This minimizes latency and power consumption.
Neuromorphic processors, unlike classic ones, do not need to access memory (or registers) and extract data from there – all information is already stored in artificial neurons. In this way, it becomes possible to process large data on peripheral devices without connecting additional computing power.
One of JAIC’s main areas of research is the creation of neuromorphic networks from so-called memristors (memory resistors). Memristors are microelectronic devices that can change their resistance depending on the amount of current that has previously passed through them. They have “memory” so are predicted to have a future as storage devices or microchips. In 2015, engineers at the University of California, Santa Barbara created the first artificial neural network based entirely on memristors.
In February 2022, JAIC was integrated into the Directorate-General for Digital and Artificial Intelligence (CDAO). This time CDAO was headed by a civilian – machine learning specialist, PhD in computer science, Professor Craig Martel.
In late 2022, The Washington Post published an overtly propagandistic article by David Ignatius discussing the use of algorithmic technology for intelligence purposes by the Ukrainian armed forces. Neither JAIC nor CDAO is mentioned in the publication, it is about software provided by the American company Palantir to the Armed Forces of Ukraine.
Palantir is an AI system for analyzing big data using neuromorphic networks, thanks to which the command of the Armed Forces of Ukraine, according to Ignatius, has similar digital maps of the theater of military operations in eastern Ukraine. But the fact that the armed forces of Ukraine do not use the developments of JAIC and CDAO shows that the Ukrainian army does not have the right to use the most advanced versions of American combat algorithms.
The algorithmic wars of the future will not be limited to intelligence. They are intended to be used, among other things, in cyber operations. Algorithms will be able to detect and exploit vulnerabilities in enemy cyber networks while driving attackers from their own networks.
At the momentacademic and political debate tends to focus on the threat of fully autonomous weapons making life-and-death decisions,” writes Lauren Gould, a researcher at Utrecht University in the Netherlands.
An example is also given. According to a report by UN experts, in the spring of 2020, during the fighting in Libya, the KARGU-2 attack quadcopter of the Turkish company STM autonomously attacked the retreating soldiers of the Libyan Field Marshal Haftar.
The drone independently targeted one of the soldiers and, without an order from the operator, fired a fatal shot. Military experts have confirmed that this is the first known case where a combat robot has independently made the decision to eliminate the enemy.
American analysts put such cases down to AI errors. “The problem is that when they (AI) make mistakes, they make mistakes in a way that no human would ever make” Arathi Prabhakar, DARPA director during Barack Obama’s presidency, liked to say.
Preparing for algorithmic wars, the United States and its allies are developing not only intelligence platforms, but also fully autonomous weapons systems that independently search for a target and independently decide on its destruction, regardless of the consequences of possible errors of such systems.
AI-controlled autonomous weapon systems are the backbone of US military strategy today. This strategy, according to the Pentagon, should ensure global leadership, primarily in the military sphere.
Translation: ES
Sign the “Peace and Sovereignty” referendum in a few seconds to open the page/
Subscribe to our YouTube channel:
and for the channel or in Telegram:
#algorithmic #wars #future
