#114 Silicon Valley's New Battlefield
In this edition of Technopolitik, Adya Madhavan writes on the military corner that AI companies have turned and the challenges military AI governance poses. Miheer Karandikar follows with a piece on the hurdles faced by India’s EV ecosystem. Finally, in this week’s curated section, Lokendra Sharma writes on nuclear fusion and its potential.
Technology has become important not just in our everyday lives, but has also become an arena for contestation among major powers including India. The Takshashila Institution has designed the 'Technopolitik: A Technology Geopolitics Survey' to understand and assess what people think about how India should navigate high-tech geopolitics. We are sure you are going to love the questions! Please take this 5-minute survey at the following link: https://bit.ly/technopolitik_survey
Technopolitik: AI for peace? Not anymore
— Adya Madhavan
Silicon Valley’s AI companies have turned a corner. After foreswearing military applications for years, players, including Open AI, Meta, and Microsoft have recently announced defence partnerships. While this may be a promising turn for governments, it can be seen as a worrying development. At a juncture when the fairness of these corporations is up for debate, the idea of them partnering with governments for defence purposes is raising alarm bells for those critical of AI. Laws and frameworks governing AI will need to be written and rewritten accordingly.
The absence of the exact details of these defence-private partnerships makes it hard for the public to understand their implications and equally hard for policymakers to legislate. Take, for instance, the examples of OpenAI and Meta’s foray into defence. OpenAI changed its policy earlier this year (that previously stood firmly against all military uses of their technology) and recently announced its partnership with defence manufacturer Anduril, stating that OpenAI would now work with the company to provide the US Military with artificial intelligence in order to defend against drone attacks according to a statement by Anduril. According to a former employee of OpenAI, the company will help the military “assess drone threats more quickly and accurately, giving operators the information they need to make better decisions while staying out of harm’s way,”. Similarly, Anthropic, the maker of Claude– a large language model,, has partnered with Amazon Web Services and Defense Contractor Palantir to enable US intelligence and defence sectors to have Claude AI at their disposal for military and intelligence purposes.
Amid concerns that large AI companies such as OpenAI do not comply with ‘Fair Use’ regulations and often use copyrighted data that they do not have permission to access—most notably by OpenAI whistleblower Suchir Balaji—these same challenges and issues spill over to the sphere of military AI, an area where governance is even more complex.
Currently, AI governance is fragmented, and governance frameworks are not yet fully developed, they still lag behind the rapid technological developments. Some of the more developed frameworks in existence, such as the EU AI Act and the OECD Principles still focus largely on civilian applications of AI and not on the military, especially not on military AI ethics and accountability for private-public partnerships. As the incorporation of AI in the military increases, AI regulation will have to be accelerated to draft laws for military AI that are ethical and enforceable.
There are arguably three broad areas where policymakers will need to focus when drafting guidelines and regulations for AI in the military:
Firstly, there are the ethical issues and questions associated with using AI in the military. Clear guidelines will need to be set to ensure human oversight, especially in areas such as the development of Lethal Autonomous Weapons Systems (LAWS) and drones for target identification and surveillance, where a lack of oversight can lead to major decisions being made by an AI system that lacks contextual understanding. The role of human oversight must be defined and mechanisms such as safeguards against some autonomous functions should be put in place to prevent misuse.
Secondly, guidelines need to be put in place to ensure accountability and transparency in the context of military AI. Accountability for AI outcomes should be defined, and this should be done for situations where unintended consequences or errors may occur as well. Additionally, accountability chains should exist where human oversight is traceable. For instance, if an autonomous drone errs and ends up striking a civilian in a combat zone, there need to be measures in place where the accountability is clearly defined, and it is easier to identify what caused the error. Moreover, explainable AI standards should be developed to ensure that human beings can interpret and understand all decisions, especially in high-stakes environments where the costs of error are high.
Finally, there is a need to ensure technical reliability and security. AI systems in the military need to be operationally reliable since errors can result in catastrophic outcomes. Additionally, these systems need to be secure from adversarial hacks. To ensure the reliability and security of AI systems, guidelines need to be in place that enforce rigorous testing and validation protocols. Additionally, resilience standards should be implemented to ensure that systems can function properly in degraded environments.
In the coming years, AI governance is likely to take on a different dimension as the technology develops. To be effective and ethical, it will have to address these various challenges and hurdles in addition to keeping up with a swiftly evolving technology.
Technomachy: Electric Horizons: India’s EV ambitions
— Miheer Karandikar
The global electric vehicle (EV) market presents a burgeoning opportunity, poised for significant growth in the coming years. India aspires to become a major player in this market, leveraging its potential to drive economic growth and achieve climate goals. This article will specifically examine the opportunities and challenges facing India's ambitions to become a major exporter of four-wheeler EVs.
Roadblocks to India's EV Export Success
China's dominance in the global EV market, particularly in exports, poses a significant challenge for India. Statistics reveal the stark difference in export volumes: China exported 1.76 million EV units in 2023, while India's share was a negligible 0.001% of the total trade. This disparity is further evident in the EU market, where China exported $20 billion worth of EVs in 2023, compared to India's much smaller figures.
Furthermore, several internal challenges hinder the Indian EV industry. The lack of adequate charging infrastructure remains a major impediment to EV adoption and exports. China has surged ahead massively in this parameter; high initial costs for Indian consumers, primarily due to expensive batteries, also present a significant barrier to entry. The underdeveloped EV supply chain in India leads to delays and increased costs, further hindering competitiveness. Despite possessing vast lithium reserves, India faces challenges in efficiently extracting and utilising them for EV battery production. A shortage of skilled labour specifically trained in EV technologies limits production scale-up and the ability to meet export demands. Concerns regarding the quality and competitiveness of Indian-made EVs in the international market are also evident from the low four-wheeler EV export figures. In the rankings of the top EV companies, no Indian company comes in the top 20.
Lessons from China's EV Export Strategy
China's success can be attributed to several key factors. Early and significant government investments in EV technology and infrastructure, including generous subsidies, tax breaks, and procurement contracts, nurtured a robust domestic EV ecosystem and drove down costs for consumers. China's mastery of battery technology, particularly LFP batteries, coupled with control over critical battery materials, has given them a significant cost advantage. Additionally, China has proven adept at adapting its EV offerings to different export markets, using varied pricing models, strategic partnerships with local businesses, and customised in-car entertainment systems to cater to regional preferences. While criticisms of China's industrial policies as "beggar-thy-neighbor" tactics exist, counterarguments suggest that certain aspects, particularly those focused on green technologies, could benefit the global economy.
Is the PLI Scheme Enough?
The Production Linked Incentive (PLI) scheme, launched in 2021 to bolster domestic manufacturing and attract investment in advanced automotive technology, has fallen short of expectations. Stringent eligibility criteria, complex documentation processes, and the resulting slow uptake of incentives have hampered the scheme's effectiveness.
While the PLI scheme represents a step in the right direction, its limitations raise questions about its sufficiency in achieving export competitiveness comparable to China. The differences in scale, scope, and implementation between the PLI scheme and China's approach are notable. The Indian government may need to consider adopting a more aggressive approach, learning from China's successful strategies.
The Way Forward: Targeted Demand-Side Interventions
Reducing import dependence by encouraging the use of domestically produced components is crucial. This could involve incentivising manufacturers to source locally, potentially addressing supply chain challenges and contributing to greater cost competitiveness. Given the EU's imposition of tariffs on Chinese EVs, exploring alternative high-growth markets that are more receptive to competitively priced EVs is recommended.
Conclusion
India needs a proactive, comprehensive, and potentially more aggressive industrial policy to achieve its EV export goals. A thriving EV export sector holds significant potential for economic growth, job creation, technological advancement, and environmental sustainability. However, bold policy decisions and strategic implementation are required to overcome the existing challenges and capitalise on the global EV market opportunity.
The Unending Promise of Nuclear Fusion
— Lokendra Sharma
Graduate students do not just study for the sake of it, but sometimes, they do it in the hope of changing the world. That is the story of a 2012 MIT graduate class on the ‘Principles of Fusion Engineering’ that tackled the problem of making nuclear fusion viable. Fast forward twelve years, Commonwealth Fusion Systems (CFS) — a company originating from the same graduate class — laid out plans for building a viable 400-megawatt nuclear fusion power plant in the US state of Virginia.
Historically, the promise of nuclear fusion has always remained ‘unending’ because fusion reactors consume more energy than they produce. Having been developed since the 1950s, they could never attain commercial viability. CFS claims to change the viability equation by deploying a breakthrough technology — a new kind of high-temperature superconducting magnet that plays the role of containing the plasma in the ‘tokamak’ reactor. CFS also claims that its reactor (called ARC) is going to be compact in terms of land requirements and help address environmental pollution and reduce carbon footprint. On the elephant in the room, that is, safety risks associated with anything nuclear, CFS says that ‘[u]nlike nuclear fission plants, fusion energy has no chance of runaway chain reactions or meltdowns, and there’s no long-lived or high-level nuclear waste.’ The investors are convinced by CFS — it has raised more than USD 2 billion in investments.
However some are not fully convinced and maintain cautious optimism. Ben Guarino in his piece for the Scientific American highlights that the International Thermonuclear Experimental Reactor in France that only intends to demonstrate that nuclear fusion is feasible (and not actually produce any commercial power) ‘is behind schedule and over budget.’ Ben Guarino also mentions how in 2021 Lockheed Martin ‘quietly shelved’ plans for building a small fusion reactor after committing to it in 2010.
But why has building and operationalising a nuclear fusion reactor proven so tough in the last seven decades? What is a tokamak? What is the current status of nuclear fusion in US, Japan, China and elsewhere? A recent 2024 paper by Sadik-Zada et al. has answers to these questions:
Sadik-Zada, E. R., Gatto, A., & Weißnicht, Y. (2024). Back to the future: Revisiting the perspectives on nuclear fusion and juxtaposition to existing energy sources. Energy, 290, 129150. https://doi.org/10.1016/j.energy.2023.129150
According to Sadik-Zada et al., ‘[a] tokamak is a doughnut-shaped magnetic confinement device that uses strong magnetic fields to confine and control the hot plasma of hydrogen isotopes (typically deuterium and tritium) to induce fusion reactions.’ Unlike a nuclear fission reactor that generates thermal energy by splitting heavy uranium into lighter elements, nuclear fusion involves generating thermal energy by fusing two lighter hydrogen isotopes to form helium. When deuterium and tritium fuse, they not only form helium, but also release four neutrons. These neutrons collide with the reactor casing or blanket, and their kinetic energy is converted into thermal energy. This thermal energy is ultimately transferred to water through heat exchangers and the resultant steam then drives the turbine which produces electricity. But deuterium and tritium do not fuse easily. They require extreme temperatures to do so. Hence, plasma is created inside the tokamak with temperatures to the tune of 150 million Kelvin. This is about 10 times higher than the temperature in the Sun’s core. To ensure that this extremely hot plasma stays away from the reactor casing, very powerful magnetic fields are used which keep the plasma essentially floating in a doughnut-shaped vacuum reactor.
Creating and then controlling this plasma has proven to be very tough. According to Sadik-Zada et al., a nuclear fusion reactor faces many challenges, including ‘engineering problems, such as creating a continued reaction within a reactor, achieving a Q-value of one or higher to meet the basic prerequisites for using nuclear fusion as an energy source, establishing a way to procure its own fuel, and finding a way to control sudden eruptions of plasma within the reactor.’ This is in addition to the primary issue that nuclear fusion reactors are currently economically unviable because of the unfavourable ratio of energy required vs energy generated. These challenges mean that nuclear fusion remains in the experimental stage worldwide.
Whether or not CFS would change this equation once and for all has to be seen.