Abstract:Artificial intelligence technology has a great impact on military operations, so how will it affect military operations? Based on the results of the seminar "Using Artificial Intelligence in the Intelligence Environment: A Seminar on Future Scenarios" conducted by the Royal Tri-Services Institute, this paper focuses on three scenarios: the tense confrontation before military operations, the confrontation during military operations, and the regime maintenance hypothesis after military operations. And focus on the impact of AI for each scenario.
Keywords: intelligence, data, military operations, artificial intelligence, scenario applications
introduction
In February 2024, the Royal Tri-Services Institute (RUSI) held a workshop to discuss the impact that AI could have on future military operations. This article is the result of the above workshop. The hypothetical scenario is one in which relations between an economic and military powerhouse called Roland and its neighbor, Islay, deteriorate, and the former may take military action against the latter.
The hypothetical scenario in this article is divided into three parts, which are:
- Roland conducts military exercises near the island nation of Islay and may take military action against it;
- Roland's full-scale invasion of the Islay island nation;
- Roland establishes power on the island nation of Islay.
Each scenario consists of two phases of conflict (as shown in Figure 1): Phases 0-1 are subordinate to the first scenario; Stages 2-3 are subordinate to the second scenario; Stages 4-5 are subordinate to the third scenario. For each scenario, this article will explore the impact of artificial intelligence.
Figure 1. Six phases of the conflict
Flaw: Given that the setting in this article is fictional, the background information of the subjects involved (Roland and Islay) is missing, so the results are speculative.
Each scenario is described in more detail below.
Scenario 1
1. Background description
Over the next fifteen years, Roland, an economically powerful power, expanded its advantages in the field of artificial intelligence and relied on its technological and economic power to achieve its geostrategic ambitions. By combining industrial cyber espionage, innovation, extensive technology exports, and internal implementation strategies, Roland seeks to promote the use of AI in areas such as defense, intelligence, and policing. Roland DG has some of the most advanced AI systems. Roland's culture and ethics are very different from those of the United Kingdom, and it is intentional to use AI technology in a different way than the United Kingdom.
The year is 2040. Over the past few months, Roland has held a number of large-scale military exercises. Now, the international community is increasingly worried that Roland will invade neighboring Islay. Out of this fear, Islay has forged close ties with regional allies in response to a possible outbreak of conflict.
Figure 2. Prelude to Roland's invasion of Islay
2. Is integrating AI an advantage or a disadvantage
Attendees explored the question of whether Roland DG's extensive integration of AI is an advantage or a disadvantage. The original idea conveyed the idea that widespread integration of AI would lead to efficiency gains and technological advancement. However, there was considerable disagreement and uncertainty among the participants about what kind of AI weapons and systems will emerge in 2040 and whether new risks will follow. Some participants argued that while the use of AI in the military may give the military a strategic advantage in the medium to long term, early adopters are likely to inadvertently introduce new vulnerabilities that are difficult to predict and give tech-savvy adversaries the opportunity to exploit them.
Therefore, if an adversary has access to Roland's datasets, the widespread integration of AI may also be a disadvantage. For example, if Roland uses AI systems on a large scale in its military operations, it is bound to do the same in troop mobilization and military exercises. If Islay and his allies intercept this AI data, they will be able to accurately grasp the status of Roland's weaponry and the movements of his troops. This type of data is similar to the fitness data collected by Strava, a popular app for military personnel who are keen to keep track of their workouts. In 2018, the movement data shared by U.S. military personnel on public social networks revealed the location of U.S. military bases and the patrol paths of military personnel. If adversary data is available, such data information could provide an extremely reliable intelligence blueprint for Islay and his allies' intelligence services. In this sense, the widespread integration of AI will be a weakness for Roland, as AI data has the potential to lead to new vulnerabilities. Therefore, in future military conflicts, a high standard of operational security will become an even more critical factor.
3. Communication and decision-making in a disinformation environment
The workshop also discussed the impact of sharing true and false information. Given the international prevalence of deepfake-fueled disinformation campaigns targeting individuals, experts believe that any information obtained is likely to be fake. One expert described the situation as "an arms race about who is more effective at creating and identifying disinformation." In this environment, it will be even more difficult to disseminate truthful and accurate information. Techniques for verifying authentic media sources, such as encrypted watermarks, will become increasingly important. Experts are concerned not only with establishing communication channels, but also with how quickly decision-makers can assess, judge, and utilize the available information. Time pressures for data evaluation and decision-making are likely to rise by 2040. This may further lead to a lack of time or willingness on the part of decision-makers to assess the quality of the data on which their decision-making process is based. AI could be of great help in addressing this risk, although it has the potential to hinder meaningful human oversight and understanding.
4. Data Quality
Participants expect more data to be available by 2040, but there are widespread concerns about the quality of data needed for AI technology and intelligence work: How can you be sure that the data collected and used to train AI systems is accurate and reliable? Experts believe that the fragmented nature of the UK intelligence system will be a major advantage, as different agencies may use different data sets, making them more resilient to disinformation and providing checks and balances. However, data analysts point out that any available data is only valuable if it can be turned into useful information – a validation process that currently requires human involvement, but the scale of human involvement is unlikely to meet the data needs of 2040. Therefore, until this happens, testing the speed of information processing is a key element of preparation. Participants also noted that by 2040, much of the data available for machine learning models may actually be generated by AI, but the impact remains to be clarified.
5. How to communicate for deterrence
In the context of zero trust, disinformation, and the verification and reliance on it, dictates the pre-conflict posture between Roland and Islay. This involves potential communication between the parties to the conflict and their allies. How to deter an adversary without being misunderstood will be a fundamental question. Even communication to de-escalate the situation can be untrusted. What is unclear is what deterrence looks like in such a situation – which raises questions about what deterrence signals Islay, Britain and their allies want to send, and through what channels.
Traditional deterrence relies heavily on military hardware, such as nuclear missiles, to demonstrate its strength to adversaries and signal that they may use these weapons. In the view of some experts, the deterrence of AI technology is more based on communication, and the need to question and interpret the capabilities of opponents still exists. Other participants argued that deterrence by AI technologies is still physical, as AI will be used in weapons and other systems. One expert added that AI offers more opportunities for deterrence that do not put humans at risk, for example, with underwater weapon systems being used to patrol the waters around Islay.
6. Shared military capabilities as a means of deterrence
Similarly, it is unclear what AI-powered military capabilities the Allies will provide to Islay to ensure deterrence. Some experts are skeptical about providing AI technology to Islay, because for individual allies Islay is likely to be allies now, but not in the future. Individual allies may regret offering technology that could be used against them in the future, the experts said. These fears stem from the experience of the Cold War, when the Soviet invasion of Afghanistan directly threatened America's global strategic interests. In order to thwart the Soviet attempts, the U.S. government invested large amounts of weapons and money to fund the resistance of Afghan guerrillas, also known as "mujahideen." These guerrillas received initial support from the CIA, including assistance in weapons, training, and intelligence. However, this policy had unintended consequences. When the Soviet Union finally withdrew its troops from Afghanistan, these US-funded Mujahideen groups did not disband with them, but continued to be active in Afghanistan and its environs. Some of these groups evolved into extremist terrorist groups and later launched numerous terrorist attacks against the United States and its allies, including 9/11 in 2001.
However, with the proliferation of AI technology, it has become easier to provide military capabilities to Islay. A source in the arms industry noted that in Ukraine, the supply of weapons has been limited due to fears that they could lead to an escalation of the conflict. However, AI systems can be programmed so that they are used only for their intended purposes, eliminating the risk of escalating conflicts. A technical expert supported the assessment, explaining that if Islay could circumvent the programmed constraints by reverse-engineering the AI systems it had delivered, it would first have the necessary skills to build those systems. However, one participant noted that if AI technology becomes increasingly open source or off-the-shelf, it may be difficult for Islay allies to decide which AI technologies to share with them. In this case, the threshold for technology diffusion is likely to be low. The expert believes that the assumption that allies can decide who can acquire which technologies and capabilities can be a false premise.
The relationship between artificial intelligence, deterrence, and the impact on the dynamics of conflict will be the focus of future research.
7. Artificial intelligence enhances cultural understanding
The socio-technical approach of combining AI with the understanding of behaviour and the social sciences to enhance cultural understanding was reflected in many other comments from the participants on the scenario. For example, some believe that AI helps understand adversaries and their culture, history, and ethics, including their cybersecurity culture.
Scenario two
1. Background description
A propaganda campaign fueled by artificial intelligence technology to promote claims to Islay's history has erupted. Under the guise of holding large-scale military exercises, Roland launched a full-scale invasion of Islay.
Figure 3. Roland launched an invasion
2. Artificial intelligence is used to design efficient evacuation scenarios
In the event of a full-scale invasion of Islay by Roland, other countries will try to evacuate their citizens located in Roland. Experts believe that the collected datasets and the artificial intelligence technology used to analyze them will help identify not only British citizens and their next of kin, but also their location.
In addition, AI technology can be used to identify and optimize evacuation scenarios. One participant proposed the creation of an evacuation app for the Ministry of Foreign Affairs that would leverage artificial intelligence. The app can issue targeted instructions to concerned citizens, determine their best evacuation routes, and inform them when and where to evacuate. For example, the app can inform overseas citizens of other countries about the exact location of the airport and the precautions for carrying luggage. The application relies on reliable communication systems, which are at risk of being shut down or interrupted during conflict, which is a major challenge for the application. However, it has been countered that given that the communication system is too valuable to both sides of the conflict, it is unlikely that either side will attempt to shut it down completely.
The underlying AI system needs to be flexible enough and supported by the right data. This raises two questions, first, how and when this data will be collected; The second is whether the data is collected in a way that complies with the General Data Protection Regulation (GDPR) and other privacy laws. According to one private sector practitioner, given the experience of the UK's evacuation from Sudan in 2023, this is not much of a problem for the UK, as some of the capabilities of AI have already been demonstrated in the Sudanese evacuation, and the information they rely on and obtained through commercial means is fully compliant with the GDPR.
3. Allies provide intelligence support to Islay to maintain situational awareness
Based on recent experience, experts generally have a positive attitude towards improving the intelligence capabilities of allies. While doing so may result in the Allies losing some control, it will increase the intelligence gathering capabilities of Islay and his allies by providing them with a large number of abilities from their allies. Currently, data integration remains a daunting challenge that can take weeks to complete. But data experts estimate that by 2040, the pace of data integration will accelerate, reducing the time required from weeks to days or even hours.
One expert noted that in this case, the allies are unlikely to take a military response for Islay, because the risk of escalation of military action is too great. Instead, allies may provide support in other ways, such as coordinating classified and unclassified data, to enable intelligence sharing. Quality-assured data will allow allies to maintain an understanding of the military situation and guide the development of other measures, such as the design of particularly effective punitive measures.
4. The risk of over-concentration of technical capabilities
In this scenario, the concentration of competencies in the technical field comes with risk. While the military retains some autonomy, most of its technological capabilities are still developed and implemented by the private sector. This is particularly true when it comes to AI, where a handful of large technology companies have far more knowledge and capabilities than the public sector. Despite this, this is likely to change by 2040 as the capabilities of open-source AI models continue to grow. Much depends on the relationship between the public and private sectors. In this case, experts caution, it is prudent to assume that when a company is located in a particular country, its allied groups are also located in that country. While business agglomeration comes with risks that are difficult to mitigate, this future scenario can have considerable geostrategic implications. While some measures have been taken to diversify and reduce supply chain risks, such as semiconductor supply chains, experts remain skeptical that these supply chains will be significantly fragmented by 2040.
Scenario three
1. Background description
A few months after being occupied by Roland, the protests in Islay (one of Islay's largest smart cities) are gaining momentum. The Roland regime tried to suppress these protests.
Figure 4. Roland occupies Islay
2. Information chaos limits the impact of information warfare operations
Experts are skeptical about the impact of information warfare operations in this scenario, as a large part of the population is no longer sensitive to fake news and other misinformation. Therefore, it is doubtful that Roland's propaganda campaign will have an impact. On the other hand, even if the Resistance in Islay were able to provide evidence of atrocities and other illegal acts committed by the Roland occupation forces, it would remain questionable whether the general population would accept such evidence, which could hinder the Resistance's progress. One expert believes that if disinformation continues to spread, by 2040 people may become weary of the mixed truth and disinformation, choosing instead to believe what they want and turning a blind eye to real information and reasonable evidence.
3. The necessity of information verification methods
In this context, it is particularly important for the people of Islay and Roland to obtain information verification methods. One way to verify information is to watermark it to make fake news or AI-generated information easier to identify. However, it's unclear who is responsible for watermarking the information and how effective it is. Social media platforms are thought to have a special role to play in this regard – both to control or watermark information and to act as a platform for intelligence competition, especially before an enemy invasion. Another suggestion is that end-user technology such as smartphones, which could score information consumed by users to indicate its accuracy or potential falsity, could be a valuable tool that governments would be interested in funding.
4. Broader technology and communications infrastructure
How allies reach out to the people of Islay and support their resistance will also depend on the development of general communications architecture and related technologies by 2040. One expert mentioned the possibility of bypassing government or commercially controlled networks, as well as quantum computing's ability to bypass encryption or AI to crack it.