When AI was involved in the Russo-Ukrainian War, eavesdropping read intelligence affected the war situation

Zhi DongXi (public number: zhidxcom)

Compile the | Yang Chang

Edit | Three Norths

According to foreign media Wired, in the Russo-Ukrainian War, an artificial intelligence (AI) tool is being used to eavesdrop on the communications of Russian soldiers on the battlefield. The tool was developed by primer, an American artificial intelligence company that primarily provides AI services to intelligence analysts.

The use of machine learning tools to analyze data such as images has been well used in the intelligence community. At the same time, the use of artificial intelligence to collect and analyze data may eventually become the core of battlefield operations. The U.S. military is investing millions of dollars in artificial intelligence software that can collect and analyze different signals in the field. Artificial intelligence has not only been applied in scientific research and work production.

The current Russo-Ukrainian war is more about the use of tanks, artillery and other weapons commonly used in the past hundred years, and with the United States and other countries continue to study the application of artificial intelligence in the military field, future wars may be more dependent on the role of artificial intelligence.

First, the soldiers retreat after being attacked, and the communication dialogue is monitored by AI

In early March, several Russian soldiers in Ukraine used unencrypted channel radio communications, and the contents of the communications were captured. The captured radio communications reveal how panicked and confused soldiers communicated with each other on retreat missions after being attacked by artillery fire:

One of the soldiers, code-named Sneg 02, said in Russian: "Vostok, I am Sneg 02." We had to turn left on the highway. ”

The soldier, code-named Vostok (Oriental), replied, "Got it." There is no need to move anymore and turn to defensive status. Complete. ”

Subsequently, a third soldier tried to contact a soldier code-named Yug 95 (South 95): "Yug 95, do you have contact with your superiors?" Tell him to be careful of artillery attacks on the highway, artillery attacks on the highway. Don't advance in columns, be careful when moving. ”

The soldier was getting more and more agitated. He continued to say to Yug 95: "Go to the radio and tell me about your situation and the location of the opposing artillery, and what weapons the opposing artillery is using." The soldier spoke again: "Give me your place, Yug 95, and answer my question." Name your location! ”

Scenes in the battlefield (Source: Wired)

While the soldiers were using unencrypted radio channels to communicate, an AI tool was eavesdropping on their conversations.

Primer's AI tool automatically captures, transcribes, translates, and analyzes conversations made by Russian soldiers using unencrypted communication channels. While it is unclear whether The Ukrainian military has also intercepted Russian communications for now, the fact that AI systems are being used on a large scale to monitor the movements of Russian troops suggests that sophisticated open-source intelligence is increasingly important in military conflicts.

Much of the russian army's transmission through unsecured communication channels has been posted on social media, and it has also been translated and analyzed. There are also clips of videos taken by people using smartphones and posts on social media that have been analyzed and interpreted in a similar way.

However, the use of natural language processing technology to analyze the content of russian military communications is still very fresh. Because deciphering intercepted communications is still labor-intensive for the Ukrainian military, analysts spend time translating information and ordering interpretations in a particular location.

The tools developed by Primer also demonstrate the value of machine learning in the analysis of intelligence information. Over the past decade, as large neural network algorithms can learn with the help of large amounts of training data, there have been significant advances in the processing power of AI technologies in image recognition, speech transcription, translation, and natural language processing.

Now, people can use artificial intelligence's off-the-shelf code and APIs to accomplish tasks such as speech transcription and face recognition, and these technologies often have a high degree of accuracy.

Faced with the superiority of the Russian army in numbers and equipment, the interception of communications is likely to have an impact on the ground forces in Ukraine.

Primer upgraded its AI tools to add 4 new features

Primer is already selling its data-trained AI algorithms. These AI algorithms can transcribe and translate phone content, including algorithms capable of extracting key terms or phrases.

Sean Gourley, Primer's chief executive, said his company's engineers modified the AI tool they sold to make it 4 new features.

The first function is to collect audio data captured from a web page output data source that is broadcast communication captured by software that uses emulated radio receiver hardware.

The second feature is noise cancellation, including constant gossip and background sounds such as playing music.

The third function is to transcribe and translate Russian speech.

The fourth function is to mark out key statements related to the situation on the battlefield. In some cases, this process involves retraining machine learning models so that colloquial terms for military vehicles or weapons in Russian soldier conversations can be identified.

Primer's end-to-end platform provides an AI infrastructure that can be quickly built, deployed, and customized to accommodate evolving intelligence and defense needs (Source: Primer)

The ability to dynamically train and retrain AI models will be a key advantage for future warfare, Gully said. He said Primer provided the tool to external parties. But Primer declined to say who it was. Gully said: "We don't say who is using this AI tool, or what users are doing with this tool. "There are several other U.S. companies that provide Ukraine with the technology, information and expertise to counter Russia.

The fact that some Russian troops use unsecured unlimited radio channels to communicate has surprised some military analysts. Peter M. Thompson, a senior fellow at the New America think tank and specializes in modern warfare, is a senior fellow at the New America think tank. Peter W. Singer said that seemed to indicate that the Russian side was under-resourced and prepared for the operation.

Singh added that in past conflicts, such as in Chechnya, the user state intercepted public communications to strike its adversaries, so the Russian side should be aware of the risks of using unsafe communications.

Calder Walton, a harvard spy historian, said the Russo-Ukrainian war had shown how valuable open source information was for intelligence agents. Facial recognition tools have been used to identify some people in conflicting videos.

The Russian-Ukrainian conflict has highlighted the importance of mining different sources of intelligence. For example, the Ukrainian military may successfully target some Russian generals by searching for gray-haired people near satellites, drones or other imaging antennas. The phones used by Russian soldiers sometimes revealed their location and mission details, as well as their emotional state.

Walton said two signals intelligence agencies, the NSA in the United States and GCHQ in the United Kingdom, are likely to have an upgraded version of the AI tool.

Third, AI is involved in international conflicts, and image and language analysis have advantages

In fact, Primer is just one of a growing number of AI intelligence analytics companies. These companies make it easier for defense agencies and private businesses to use such AI technologies. The involvement of private technology companies such as those providing satellite communications and imaging in the Russo-Ukrainian War has sparked reflections about the rights these companies have been given and how to get involved in international conflicts.

Leveraging open source intelligence data means sifting through large amounts of information. Emily Harding, a senior fellow at the Center for Strategic and International Studies, a policy research nonprofit, says no one can handle the vast amounts of open source intelligence. Harding wrote a report in January 2022 on the use of ARTIFICIAL intelligence technology for open source intelligence data mining. Harding noted that the intelligence community is doing better at using machine learning tools for image analysis. She added that Primer is known for its ability to parse languages.

New advances in artificial intelligence could make the technology a more powerful tool for performing text and speech analysis. In recent years, thanks to the emergence of a large machine learning model called Transformer, AI has been able to use text to summarize and answer questions.

This type of model is able to better understand input, such as a long list of words in a sentence. Based on the Transformer model, artificial intelligence model programs that can "write" coherent news articles and even write computer code that can perform a given task have been developed.

Still, Harding noted that the intelligence community will also have to "contend" with the same problems that plague the deployment of AI in other areas, such as algorithmic biases caused by poor quality or unrepresentative training data. Because machine learning algorithms work in an opaque way, she says, intelligence officers need to find ways to build credibility in the conclusions reached by these AI programs. Of course, mistranslated communications can have fatal consequences on the battlefield, such as putting soldiers in dangerous situations or misleading missile attacks.

The use of artificial intelligence to collect and analyze data could eventually become the core of battlefield operations. The U.S. military is investing millions of dollars in artificial intelligence software that can collect and analyze disparate signals in the field.

A U.S. Army program called the Tactical Intelligence Targeting Access Node proposes creating a ground station that can capture and extract intelligence from many different battlefield sensors and data sources.

If the current Russo-Ukrainian war relies more on tactics of the past century, such as tanks and artillery, then future wars that the United States and other countries are preparing may rely heavily on new technologies such as artificial intelligence. These new technologies can improve critical information faster and help win faster. At the same time, some experts have stated that when using artificial intelligence on the battlefield, deceiving and misleading algorithms are equally important.

Primer CEO Gully said their philosophy on AI and defensive AI is that no matter what algorithms people apply to war, those algorithms won't be the ones that people end up getting.

Conclusion: When AI is involved in war, it needs to be vigilant

In the Russo-Ukrainian War, the smart technology or products used may have gone far beyond what Primer provided. In terms of actual results, Primer's artificial intelligence tools can already obtain and analyze call information from unsecured communication channels.

With the development of artificial intelligence technology, the capabilities they embody are getting stronger and stronger, and how to use artificial intelligence well may be a problem that both industry and academia need to think about together.

Source: Wired