A summary of Rajiv ji’s interview with Indian Military experts on the State of Indian Military and AI.
Air Vice Marshal Bahadur questions General Pannu and Rajiv Malhotra in light of the latter’s recent book “Artificial Intelligence and the Future of Power”. Some of the key points of this very interesting discussion on how AI impacts defense and national security are presented.
AVM Bahadur Manmohan
AVM Bahadur Manmohan introduces the topic, describing artificial intelligence as a branch of computer science that’s concerned with the development of machines with the capability of completing tasks that typically require human intelligence. This capability has greatly increased over the last four decades because of the ever-increasing availability of compute power that is essential for machine learning. Also, while both civilians and the military have been researching the subject, civilians have made greater advancements. So their work is currently being adopted by the military with adaptations.
He sees China as the leader in AI R&D, which it is currently using for better decision-making as well as in autonomous military vehicles. Russia is very serious about the subject, focusing on both robotics and AI. Putin has been known to say, “Whoever becomes the leader in this field will rule the world.”
The U.S. national defense strategy released in 2018 says that its focus is on winning the wars of the future. American venture capitalists (VCs) invested $8 billion in R&D in 2018 alone. A congressional research paper says that the Department of Defense’s (DoD) unclassified investment in AI is up from $600 million in 2016 to $2.5 billion in 2021 (it has 600 active AI projects).
He feels that AI will have the broadest application in modern warfare, facilitating autonomous operations, speeding up military decision making, increasing the speed and scale of military actions, in ISR (Intelligence, Surveillance, Reconnaissance), in logistics, in cyber operations, etc.
But because of its nature, he feels that it is also vulnerable to unique forms of manipulation. And because it is such a powerful tool that is increasingly coming under the control of people and also, machines, the government must have suitable policy on its use and must make R&D investments in its development.
Without the government’s involvement, this powerful tool can be used mischievously for things like spreading fake news that could have far-reaching consequences. It could also be used for information warfare.
He also mentions the UN committee in Geneva that looks at lethal autonomous weapon systems.
Rajiv Malhotra begins by explaining his interest in the topic of AI. As a computer science student, he had worked on AI in its early stages. He is now looking at the issue again after more than 4 decades and studying it in terms of its implications on society, politics, economics, unemployment and the military.
Rajiv mentions that he has studied in depth both the U.S. and China and their recent conflict, which is far more than a trade war. The Trump administration, in particular, was very concerned that China had stolen secrets in advanced technology (includes robotics, drones, nano-technology, aerospace, etc) from the U.S. The U.S. fears the repercussions of the theft since AI acts as a force multiplier, turbo-charging research in many cutting-edge technologies.
Rajiv explains how he has gone through different reports on AI including an important one by Niti Aayog and it is evident that India is many years behind both China and the U.S. in terms of AI adoption. However, the government and particularly the military have started taking the matter seriously.
He says that AI is a game changer. It is much more than technological evolution. It’s scale of impact will be comparable to the industrial revolution and electrification of factories, redefining the haves and have-nots. In this new era, it will be harder to separate the military, political and economic aspects of a country because more economically powerful societies can invest more in the technology and assume control over others. Additionally, a more cohesive society will achieve more than one that has to devote resources to manage internal conflicts.
He highlights India’s weak position with some data:PricewaterhouseCoopers forecasts that by 2030, the incremental AI economy to global GDP will be $16 trillion. China is expected to have a $7 trillion share, followed by U.S., Europe, Southern Europe, Asia, Japan, South Korea etc. India, clubbed in Rest-of-the-world just has a mere $1.2 trillion share.
He also quotes some UN statistics: the U.S., China and Japan have 80% of AI patent filings today, US and China have 75% of block chain patents and 75% of all cloud computing market share (the 70 largest digital platform companies have 90% of their market capitalization in the U.S. and China).
As apparent from the data, he says that both the U.S. and China have got in the game much sooner and have therefore mad a lot of headway. China in particular has bet its future position in the world on AI.
He further elaborates on how the U.S. went about it by initially letting its corporations like Google, Facebook, Microsoft, Netflix, Amazon, etc develop AI with defense being involved on a much smaller scale. The U.S. leverages what he calls the military-industrial-academic complex (General Eisenhower coined the term military-industrial complex to describe a close alliance between the two groups, with academic also joining in later). This has created a huge R&D pipeline and led to notable advancements and innovations.
As an example of how closely these groups function, he says that recently, the U.S. appointed former Google CEO Eric Schmidt to head its national security effort in AI. Their official charter says that the effort is to counter the rise of China in AI.
He says that the U.S. DoD had always played an important role in cutting edge technologies. The Defense Advanced Research Project Agency (DARPA) first developed the Internet, they played a pioneering role in driverless cars (organized a 100-200 mile race in the Nevada desert that companies participated in, and continued it for many years until there were some good success stories). Defense contractors like Lockheed Martin, Boeing, General Dynamics, Northrop Grumman, Raytheon and more recently, tech companies like Google and Microsoft are part of the defense establishment.
The U.S. defense has many programs running. One is the development of driverless submarines being operated from a central place somewhere.
Adopting the same model, Chinese companies Alibaba, Tencent, Baidu, Huawei and 100s under them are closely aligned with the Peoples Liberation Army. The private sector brings academics who bring students, thus building a big pipeline.
India doesn’t have anything like this and it looks vulnerable because Pakistan has likely acquired this AI technology from China.
While India produces the largest number of STEM (science, technology, mathematics) graduates, most are doing mundane things. The brightest go abroad or work for Indian branches of multinationals like Google and Microsoft. There are also a lot of Chinese VCs, which invested around $5 billion in India through front companies in Europe and Singapore.
The number of Indian researchers in AI is 10th largest in the world, the number of Indians quoted and cited all over the world on AI is 13th largest (excluding those working for other countries/companies/ecosystem).
Indian companies like Tata, Wipro, TCS, etc are working for others. They’re in the middle of and managing AI for clients and governments but not reinvesting sufficient profits from the labor arbitrage into the future. So we are building intellectual property for others then licensing it back. He says that the Niti Aayog report that was very critical of India’s inability to use its talent had nailed the problem.
India’s human resource is a big issue, and there also many questions on where we are in AI, how much we are importing, what big data we have, our funding, strategy and ecosystem.
He suggests that we need an organization of the scale of ISRO, Bhabha Atomic Research, etc for AI, with somebody like Homi Bhabha, Vikram Sarabhai or Satish Dhawan to lead the effort. The separate academic, military and industrial silo system is not going to cut it, considering the amount we have to catch up, being 10 years behind China.
He says that importing AI means that not only are you dependent, but you’re also one step behind. Because you’ve invested in that platform, it’s hard to move away. And if you have to stay, you must keep going back for upgrades which won’t be cheap. Additionally, you won’t ever have what that partner of yours doesn’t have, and any partner today could also be an enemy tomorrow. This is the process of becoming what he calls a digital colony.
In this new world, the U.S. and China are colonizers; China has already colonized Pakistan and large parts of Africa using this technology (surveillance, sensors, facial recognition, tracking, etc).
India can be free only if it has its own systems, developed through joint collaborations.
India’s concerns about China have led it to take steps in quantum computing. TIFR (Tata Institute of Fundamental Research) has started a program on this.
Lieutenant General PJS Pannu
Lieutenant General PJS Pannu starts by outlining India’s boundaries that it is required to protect: 7,000 km of coast line and close to double that in borders that it must protect.
He says that some of these borders are hotly contested. The PLA attacked us last year using conventional means.
He explains that military systems today have moved from command, control and communication to command, control, communication, computers and cyber. The party observing, correctly deciphering, deciding and acting faster than the other will win any confrontation. So the one using the new and more efficient tools will have an edge over the other. Moreover, if you take an incorrect decision, there’s no scope of changing that decision. But if you take more time to decide, you could lose out to the enemy.
AI powered systems are limited by the imagination/scope of the people manufacturing and controlling them. So if an irresponsible person gains in control of them, there can be a disaster. Only systems supported by the government and in accordance with proper rules and regulations should be used.
There is also the real concern of your own system being hijacked. This could happen in a number of ways. The input itself could be incorrect, manipulated or unavailable, leading to incorrect decisions. With AR/VR (actual/virtual reality) technology, you can be tricked into believing that you’re dealing with a human when you may actually be dealing with a machine. Also the feedback you receive after hitting a target (for example) telling you that it had been destroyed, could be incorrect. This could happen either because the system was n incorrect image was picked up or because the system was hacked. He narrates a personal example to illustrate: a message came that the drone was returning home when it had already crashed (achieved by manipulating information either because algorithms were shared by the manufacturer or stolen by the enemy)
He says that the surveillance and reconnaissance function now has a third part: robotics. The AI system will receive information, process it and act on it via a robot. Since the destruction in this case will be automatic, there is no second chance to change the decision, the robot will not check back before neutralizing the target using drone, aircraft or tank. Also, unmanned vehicles in this system can become multidimensional, moving in the air, undersea or on land to get the job done. In coastal areas, amphibious vehicles can come from the ocean, fight and go back to base.
A self-learning machine can become devastating if it can’t be properly controlled.
China has gone into quantum technology, which can either secure your systems very well or break codes faster than you can program them. It can access anything, hack any system, or alter any programming to render your system ineffective just as you were getting ready to act/push the button.
India is using systems manufactured by foreigners or Indian vendors who haven’t upgraded to the latest technology. So the army at Siachen is tired, the equipment is overused, and even after all that security remains an issue. We must be ahead of the others (if industry is now at 4.0, military must think 4.5). He supports the military-industrial complex.
He voices a huge concern about importing AI systems: either the data would be irrelevant because it wouldn’t relate to you, or because the data hadn’t been converted into algorithm. So you have to feed in your data into that system and allow somebody to program it for you. And you would be dependent on that person for all your security. We already rely on others for chips and we don’t know how the data on that meshes with our data.
He suggests that lethal autonomous weapons be developed in conjunction with the new generation on the technology side and veterans/retired people with fighting experience on the other. That way the weapon wouldn’t act prematurely. Chances of accidents are very high today.