#92 The Beginning of China's AI Command?
China has developed an AI Commander– What is it for?; Keep an eye on Agnibaan
Today, in their inaugural post for Technopolitik, Adya Madhavan shines a light on the capabilities of China’s newly revealed AI military commander. Ashwin Prasad follows with a short primer on why watchers of the Indian space sector should watch the progress of Agnikul’s Agnibaan rockets.
If you like the newsletter, you will love to read our in-depth research and analysis at https://takshashila.org.in/high-tech-geopolitics.
Also, we are hiring! If you are passionate about working on emerging areas of contention at the intersection of technology and international relations, check out the Staff Research Analyst position with Takshashila’s High-Tech Geopolitics programme here.
Cyberpolitik: China has developed an AI Commander– What is it for?
— Adya Madhavan
Chinese scientists recently revealed that they have created a virtual AI military commander that is already participating in war game simulations. This development has led to a flurry of speculations about potential use cases of military AI, possible ethical issues, and regulatory mechanisms that need to be implemented when deploying them.
Currently housed at the Joint Operations College in the Hebei Province, this virtual commander brings with it a host of challenges and possibilities. As of now, the commander’s sole purpose seems to be to participate in large-scale military war games that simulate conflict. Reportedly, the AI commander reacts to events in real-time, “learning and mimicking the thought patterns of a real military commander.”
One of the most significant logistical issues with war games at military institutes is often the unavailability of human actors at the commander level. In addition to having official duties to perform, there are simply not as many PLA officials above a certain level of seniority who are available to participate in simulation war games. The AI commander has the potential to fill that gap.
Most developments in AI (especially military AI) raise questions and concerns about ethics and accountability. However, China maintains that ‘The Party commands the gun’—or the computer in this case. Currently, Beijing claims it only intends to use AI in tandem with human actors and not in any independent capacity.
Instead of training its AI commander not to succumb to human fallacies and weaknesses, researchers trained it on various human commanders' personalities, decisions, and combat styles. To simulate them accurately in war games, they added an essential aspect of human cognition: memory. The AI commander's knowledge base has a fixed capacity beyond which it ‘forgets’ older information when exposed to new data beyond a specific limit. This feature is particularly fascinating because one of the most useful features of an AI system is the potential for its memory to far exceed that of humans. This is what enables AI to excel at games of strategy like Chess and Go, thinking thousands of moves ahead and considering endless possible outcomes.
When it comes to military AI, most speculation regarding capabilities mainly refers to AI that could surpass human decision-making, relying on its vast capacity to retain data. China’s AI commander, on the other hand, seemingly has the aptitude to make data-driven decisions without being hampered by the longer decision-making time that humans tend to have while still being constrained by a ‘finite’ memory.
For the purpose which China currently claims to have designed its AI commander– participating in war games in place of human leaders, the in-built deficit enables it to train those participating in war games with the advantages of quicker, more empirical decisions without teaching them unlikely scenarios that human commanders would likely not authorise. Since China maintains that this AI will not be making any actual autonomous tactical decisions in battle, training AI to function as a commander and not surpass a commander seems like an effective way to ensure smooth and efficient training without actually utilising the time of senior PLA officials to help extensively train it. While AI may continue to be developed for autonomous deployment on and off the battlefield, Beijing’s current method of deploying AI gives food for thought on the ways in which AI can be utilised with a combination of features and human functions that need not necessarily be close to its peak performance but can prove more useful in narrower, specific contexts.
Additionally, the role of command extends beyond solely making tactical decisions. As long as at least significant portions of warfare involve human actors, the need for leadership, morale-building, and other activities that require human interactions will continue to be a large part of the role.
Many have marked this development as the beginning of AI’s foray into making tactical decisions on the battlefield, but it can be argued that it is still a long way off. Artificial Intelligence excels at processing large amounts of data and information to make data-driven decisions. In its current capacity, it does not possess the ability to use intuition without complete information. While it excels at coming up with the best possible outcome using the data it has been trained on and has access to, incomplete information is the rule rather than the exception in warfare.
Without information, human decision-making relies on instinct, sometimes emotionally reliant choices, and an understanding of the context based on (difficult-to-abstract) personal lived experiences. AI relies entirely on the data it has been trained on and, for now, cannot generalise beyond what it has been exposed to without being given additional information or inputs from which to learn. Additionally, the quality of the data used for its training determines the calibre of its responses and recommendations. This means that incomplete, faulty, or biased information will affect AI’s ability to make good “intuitive” decisions when faced with unknown/unforeseen scenarios.
Until artificial intelligence can perform abductive reasoning and devise ‘creative’ solutions, much like with Beijing’s virtual commander, it is likely to be deployed only for decision support or training purposes, not autonomous decision-making in warfare.
India is set to host the Quad Leaders' Summit in 2024. Subscribe to Takshashila's Quad Bulletin, a fortnightly newsletter that tracks the Quad's activities through the Indo-Pacific.
Your weekly dose of All Things China, with an upcoming particular focus on Chinese discourses on defence, foreign policy, tech, and India, awaits you in the Eye on China newsletter!
The Takshashila Geospatial Bulletin is a monthly dispatch of Geospatial insights for India’s strategic affairs. Subscribe now!
Antariksh Matters: Keep an eye on Agnibaan
— Ashwin Prasad
**The ideas in this piece first appeared in this article on Firstpost.**
Up until recently, ISRO made every single Indian rocket that could take payloads into space. Two Indian startups have now begun making launch vehicles of their own. Skyroot, a Hyderabad-based startup, tested its suborbital-sounding rocket in 2022. Following suit, Agnikul Cosmos, based in Chennai, has tested their slightly more advanced suborbital rocket a few weeks ago.
Agnikul has faced its fair share of hiccups in the process. Multiple prior attempts at launch were cancelled, (some in the last minute) before finally tasting success in this fifth attempt. Space launch technology is not easy and comes with high stakes and high risks. It is good to see an Indian company persevere and succeed, especially due to some of the novel technologies Agnikul brings into the mix.
The Engine
The rocket, Agnibaan, uses a 3D-printed semi-cryogenic engine. Semi-cryogenic engines are among the most popular rocket engines used at the present. Cryogenic and semi-cryogenic engines boast higher energy density and specific impulse than traditional solid-fuelled and liquid-fuelled engines. Specific impulse is an indication of how efficiently the engine converts propellant to thrust. In simpler words, it is the rocket's 'mileage'. Compared to solid-fuelled variants, semi-cryogenic engines allow for higher variability of thrust—and the ability to throttle. There are other advantages, too. Compared to liquid-fuelled ones, they also have a higher thrust-to-weight ratio. Also, propellant storage and handling are simpler in semi-cryogenic rockets. ISRO also plans to use a semi-cryogenic stage in its LVM3 to boost its payload capacity.
The Mobility
Agnibaan launched from a mobile launch platform - Dhanush. Its transportability means that, in the future, the launch site can go to the payload rather than the payload going to the launch site. This can do wonders for last-mile launch delivery and also offers some strategic advantages. The laws of physics dictate that all spacecraft pass through the antipodal point after launch before completing their first orbit. The antipodal point is a precise point on the opposite side of the globe from where the launch occurred. Thus, if adversaries can potentially blockade the antipodal points of a country's spaceports, they can cut off the launch vehicle’s access to orbit. Transportability of the launch site can offer a way to mitigate this vulnerability.
I am curious to see how Agnikul will deliver on its potential in the coming months. The next obvious milestone would be an orbital test flight. While they aim for higher altitudes, I would like to see them aim for higher capacities as well. Each of the Agnikul engines delivers 25 kN of thrust. This should scale up so that Agnibaan's future variants can carry heavier cargo. Until our technology advances enough for us to launch payloads from space itself, we are relegated to taking everything we need, with us from Earth. Thus, heavier launch capabilities are unlikely to go out of fashion for the foreseeable future.
What We're Reading (or Listening to)
[Takshashila Discussion Document] China’s ‘1+N’ Policy Framework across Provincial Governments: Implementation, Progress and Challenges, by Rakshith Shetty
[Opinion] Putin’s North Korea visit making China uneasy, by Manoj Kewalramani and Aadit Pareek
[Video] Young Scholars Analyse China's Politics, Economy And The Military On SNG Roundtable ft. Anushka Saxena and Amit Kumar
[Takshashila Blog] Fishing For Power: Tensions in the South China Sea, by Ananya Venkateswaran