Can technical experts influence national AI policy? The case of Homi Bhabha and the Indian nuclear program

Note: I wrote this paper as a 2022 Summer Research Affiliate at the Cambridge Existential Risks Initiative, kindly mentored by Dr Waqar Zaidi. I presented this paper at the Stanford Existential Risks Conference, where I was the youngest presenter and the only one who hadn’t yet started college.

Abstract

Historically, the decision to pursue a nuclear weapons program has been undertaken by heads of state who were often elected officials or career bureaucrats. It is unusual for a scientist to hold sway over this decision, as nuclear weapons alter a nation’s grand strategy, can be prohibitively expensive to develop, especially for developing countries, and raise complex ethical and legal issues that a technical expert likely lacks the authority and expertise to solve.

Hence, the case of the physicist Homi Bhabha stands out.

Bhabha lobbied the Indian prime minister Jawaharlal Nehru into providing the Atomic Energy Commission near-total autonomy and secrecy in its operations. While ostensibly researching only peaceful uses of atomic energy, Bhabha adopted a strategy that would allow India to pursue nuclear weapons capability if it wished to in the future. For example, Bhabha’s nuclear materials acquisition program emphasized the maximization of plutonium output, a dual-use element crucial to developing nuclear weapons. After Nehru’s death, Prime Minister Lal Bahadur Shastri was forced to allow Bhabha to continue nuclear weapons capability research despite his personal moral revulsion to nuclear weapons, owing to Bhabha’s influence over the media and his ability to leverage a small but vocal pro-bomb lobby in the parliament.

What explains Bhabha’s outsized impact? Early in his career, he gained access to a key decision-maker, Nehru, who could override conventional political and bureaucratic processes to grant Bhabha’s research program autonomy and secrecy. This was due to Bhabha’s strategic alignment (both believed that nuclear energy and weapons would bolster India’s economic development and security) and shared cultural background with Nehru, Bhabha’s preeminence in India’s nuclear physics establishment, his political entrepreneurship and savvy, and lastly, his close connections with foreign scientists who could provide technical expertise and raw materials necessary for India’s nascent nuclear program.

I also analyze Bhabha’s contemporaries to arrive at a more nuanced view of the importance of different factors. For example, the case of one of Bhabha’s contemporaries, Meghnad Saha, is illustrative of the importance of strategic alignment with a key decision-maker: while Saha was a leading nuclear physicist and showed political entrepreneurship, he disagreed with Nehru’s pet nuclear policy. As a result, Saha couldn’t gain the decision-making power granted to Bhabha.

This case study suggests that a technical expert can exert outsized influence over the development of strong artificial intelligence, which raises the same issues as nuclear weapons—expense, ripple effects on grand strategy, and legal and ethical concerns—if they gain access to an influential decision-maker via the factors enumerated in the paper.

Sources

I extensively relied on George Perkovich’s India’s Nuclear Bomb, Robert Anderson’s Nucleus and Nation, and Itty Abraham’s The Making of the Atomic Bomb.

As a first introduction to Indian nuclear policy, I recommend Harsh V Pant and Yogesh Joshi’s Indian Nuclear Policy in the Oxford India Short Introductions series.

As a first introduction to the life and work of Homi Bhabha, I recommend his Wikipedia page! I extensively edited it, nearly quintupling it in size, while working on this paper.

Timeline

To provide useful background before diving into Bhabha’s impact on Indian nuclear policy, here’s a timeline of the program’s milestones:

1939: Bhabha returns to India from his theoretical physics laboratory in Cambridge for an annual vacation; World War II breaks out, forcing him to stay.

1940: Reluctantly, he becomes a Reader at the Indian Institute of Science, Bangalore; he quickly rises up its ranks, becoming India’s first Adams Prize winner, and an elected Fellow of the Royal Society at the unusually young age of 31.

1945: Bhabha establishes the Tata Institute of Fundamental Research for basic physics research with a private grant, and becomes the chairman of the Atomic Energy Committee which disburses government money to research projects.

1948: Bhabha writes to the Indian Prime Minister Jawaharlal Nehru a proposal to establish “a very small and high powered body composed of say, three people with executive power” that would administer the development of nuclear energy in the country. Bhabha’s proposal is accepted, and he is made Chairman of the Atomic Energy Commission.

1952: Under Bhabha’s directorship, the public company Indian Rare Earths Limited extracts radioactive thorium from the monazite sands of southern India’s beaches.

1954: Bhabha announces the three-stage plan for nuclear energy tailored to India’s unique geographic situation—its abundance of thorium and paucity of uranium:

• Purchase naturally occurring uranium-235 from abroad. Churn it through reactors that transmute uranium to plutonium while generating energy.
• Put this transmuted plutonium through another reactor which transmutes it to uranium-233 while generating energy.
• Uranium-233 is a special fissile material; put it into a breeder reactor with a dash of thorium, and more uranium-233 comes out than went in—while generating energy. Repeat this step for energy output unconstrained by uranium supply.

However, the plutonium produced after the first stage could in principle be used to create nuclear bombs.

1958: The Phoenix plant which can extract plutonium from “spent fuel” produced by natural uranium reactors completed.

1960: The Canada India Reactor Utility Service (CIRUS), which takes in natural uranium and belches out plutonium-rich spent fuel, is completed. Paired with Phoenix, it produces India’s first weapons-grade plutonium, which would later be used in India’s first nuclear explosive.

1964: China explode their nuclear bomb. Bhabha fuels a pro-bomb lobby in the media and the Parliament to force Nehru’s successor, Lal Bahadur Shastri, into permitting the AEC to carry out research in nuclear explosives.

1966: Shastri dies on a diplomatic mission; less than two weeks later, Bhabha dies in a plane crash.

1968: India, unhappy with the terms of the Nuclear Nonproliferation Treaty, (privately, internally) decides to build a nuclear explosive.

1974: India tests its first nuclear explosive.

Bhabha’s impacts

With this background in place, we are ready to see how Bhabha influenced key aspects of India’s nuclear policy.

Securing autonomy and secrecy

The 1948 Atomic Energy Act gave the Atomic Energy Commission authority and secrecy almost completely unchecked by the legislature.

Anderson gives an anecdote to illustrate this: when Nehru’s Finance Minister was troubled by the eye-watering sums being spent by the AEC and demanded to see an expenses report, Nehru gave him a summary of the AEC’s report. Nehru forbade Deshmukh from showing even this summary to anyone else, and said that if Deshmukh wished to see the whole report, he should not make copies of it. The only other person who got a copy was Gulzarilal Nanda, the Home Minister (the functional equivalent of the US Vice President). Nehru’s letter to Nanda attached with the report noted: “I do not propose to send these papers to anyone else.”¹

Even the highest-level members of Nehru’s cabinet weren’t privy to the AEC’s work and finances. As Nehru briefed the Parliament in a rare session on nuclear energy, the Department of Atomic Energy’s budget “increased twelve-fold” between 1954 and 1956, yet “nobody in the Government of India—neither the Finance Ministry nor any other Ministry—anxious as we are to have economy to save money, has ever refused any urgent demand of the department.” The DAE wasn’t spending paltry sums either: in the late 1950s, almost 30% of all governmental investment in R\&D went to the AEC.² Bhabha had almost unfettered access to these funds, as the AEC dealt directly with Nehru.

Why did the AEC get so much power? As a document that was released into the public domain as recently as March 2012 shows, it is because Bhabha himself insisted on such secrecy and autonomy.

On 26 April 1948, Bhabha sent Nehru a “Note on the Organisation of Atomic Research in India”; he said that the first goal for developing nuclear energy should be the setting up of a reactor. The first reactors had to be powered by imported uranium; however, no nuclear power would be willing to share its technical know-how and fuel unless it was guaranteed secrecy by India. For this reason, Bhabha recommended that “the development of atomic energy should be entrusted to a very small and high-powered body composed of say three people with executive power, and answerable directly to the Prime Minister without any intervening link.” For brevity, Bhabha recommended, “this body may be referred to as the Atomic Energy Commission.” The final AEC resembled Bhabha’s proposal almost to the word.

He also showed why the existing authority on atomic energy in India, the board of Research on Atomic Energy of the influential Council of Scientific and Industrial Research, was not up to the task: 28 members strong, it was too large to manage these “secret matters”.³

Creating a nuclear weapons option

The first of Bhabha’s proposed three stages was the production of weapons-grade plutonium. With CIRUS and Phoenix, Bhabha had established a steady supply of this element that the Indian nuclear establishment leveraged for the 1974 detonation.

Perhaps more importantly, Bhabha lobbied the International Atomic Energy Agency to weaken its safeguards. The Agency’s draft statute had required that fissile materials like plutonium would be deposited with the Agency, which would then ration out quantities it considered sufficient for peaceful uses to the states.

Bhabha vehemently opposed this, declaring it “the inalienable right of States to produce and hold the fissionable material required for the peaceful power programmes”. In the end, he won; the final statute only contained easily-bypassable safeguards on fissile material stock and reactors, and didn’t recommend Agency control of these resources.⁴

Finally, Bhabha hired plutonium metallurgists with the stated goal of engineering plutonium fuel for the breeder reactors of stage two in Bhabha’s plan. These breeder reactors weren’t even a distant possibility at this point: as of writing, the first breeder reactor is still being built in India, and Bhabha’s three-stage plan, which the Indian government formally adopted in 1958, is expected to complete by 2050.

These metallurgists would prove valuable soon, though; they produced the core of India’s first nuclear explosive in 1974.

Perkovich also claims that the Phoenix reactor had capacity far outstripping current or planned reactors, and more than could be put to economic use—suggesting military use.⁵

Intended or not, Bhabha’s quixotic three-stage nuclear plan served as a solid foundation for a weapons program in the future. It’s not clear a scientist other than Bhabha would have pursued the thorium fuel cycle if put in charge of India’s nuclear program, and consequently produced a useful spate of weapons-grade plutonium.

Lobbying to create nuclear explosives

Lal Bahadur Shastri succeeded Nehru as Prime Minister after his death in 1962. A Gandhian,⁶ Shastri had a strong moral aversion to nuclear weapons. Any concessions Shastri made to nuclear explosives research would have to be hard-earned.

Luckily, the 1964 Chinese nuclear tests presented just such an opportunity for Bhabha. India had just lost a war to China; the Communist giant possessing nuclear weapons alarmed the Indian public.

So, Bhabha used two separate occasions to create a public impression that nuclear weapons could be created quickly and cheaply. At a London address, Bhabha promised that India could build a nuclear weapon within eighteen months of a decision to do so.⁷ At an All India Radio broadcast, he said that a nuclear explosive the order of the one dropped on Hiroshima could be produced for \$300,000.⁸

Both these figures were mistaken. It took India six years to build a nuclear bomb after it decided to in 1968 (or, four times as long as Bhabha had estimated); Raja Ramanna, who took on AEC’s reins after Bhabha, said: “I don’t think it would have been possible to do what Bhabha said—build a device in 18 months. A crash program could have been done, I suppose, but it would have been very expensive.”⁹ Bhabha’s low cost estimate ignored startup expenses—on reactors, reprocessing plants and other manufacturing infrastructure.¹⁰ To put into perspective how low the \$300,000 number is, consider that the Manhattan Project cost over \$2 billion, making the average cost per US bomb more than \$500 million.¹¹

However, Bhabha’s numbers caught on in public. A newspaper poll found that the Indian elite now took “for granted” Bhabha’s eighteen-month timeline. Pro-bomb advocates in the Hindu nationalist Jana Sangh and the leftist Samyukta Socialist Party latched on to Bhabha’s incredibly low cost estimates as support for a potential nuclear weapons program, even after Shastri produced a more reasonable \$85-105 million/bomb figure after consulting British experts, and the American scientist Christopher Hohenhemser stepped in to give a >\$70 million figure for a nuclear program in its first year of operation.

Having caused this public stir, behind the scenes, Bhabha launched a campaign to build nuclear weapons in the cabinet. A six-hour discussion culminated in all but three cabinet ministers—one of them being Shastri—endorsing a nuclear weapon-building program. Shastri was forced to authorize Bhabha to “come up with estimate [sic] of what was involved in India’s attempting an underground ‘explosion’.”¹²

Shastri was already being perceived as weak by senior officials of the Congress for his poor handling of the ongoing food crisis. He was opposed to a weapons program eating into the development budget. However, this perception of being weak is also what forced him to authorize the bomb development program.

So, Shastri reached a compromise, whereby he allowed Bhabha in the Parliament to pursue “peaceful” nuclear explosives capability. This was all that Bhabha needed, anyway; since what separates a “peaceful” device from a “weapon” is only intent, not technical specifications, Bhabha could work with what Shastri had given him to build India’s nuclear weapons option.

In exchange, Bhabha admitted publicly that his cost estimates for the nuclear explosion were too low, since they stemmed from an American study on already-established nuclear energy infrastructure. He also reiterated his commitment to peaceful uses of nuclear energy only.

Raj Ramanna and Homi Sethna, both former chairmen of the AEC, have since said in interviews that Shastri’s reversal on the no-explosives policy that day must have come from pressure put on by Bhabha.¹³

Shastri’s goal seems to have been to win Bhabha over to his side. With Bhabha ceasing to issue public statements suggesting a nuclear weapons program, and the low atomic weapons program cost nullified, the pro-bomb lobby died down. Bhabha continued chipping away at the Subterranean Nuclear Explosion for Peaceful Purposes (SNEPP) program.

The factors determining Bhabha’s success

In this section, I’ll speculate on the factors that led to Bhabha’s success in influencing Indian nuclear policy. For almost every factor, I’ll include a negative case study: an Indian scientist poised in nearly every way to achieve Bhabha’s levels of influence, but who didn’t because he lacked one of the key factors that led to Bhabha’s success.

Preeminence in nuclear physics research

Few politicians had the expertise to critique Bhabha’s three-stage nuclear program; almost the only exception was Meghnad Saha, the scientist-turned-parliamentarian,¹⁴ whose case I discuss later in the paper.

Perkovich writes that “India’s scientists for the most part lacked reasons or standing to critique Bhabha’s plans or the powerful Department of Atomic Energy”.¹⁵ He notes that even Shastri took care not to attack Bhabha by name in the Parliament, since Shastri’s opponents would invoke Bhabha’s scientific distinction, and his honor. Instead, he only obliquely pointed out that scientists shouldn’t make policy prescriptions.¹⁶

Negative case studies

Shanti Swaroop Bhatnagar

The head of the influential Council for Scientific and Industrial Research (CSIR), Order of the British Empire, Fellow of the Royal Society, and a knight, Shanti Swaroop Bhatnagar was higher up the political ladder than Bhabha was in the mid-1940s, when the Atomic Energy Committee was established as a subcommittee of the CSIR. Bhatnagar also had foreign connections, as a senior officer of the London-based Society of Chemical Industry; his knighthood gave him “subtle social advantage” among his British friends.¹⁷

Almost by default, then, Bhatnagar became the subcommittee’s chairman when it formed. However, by 1946, as the minutes of meetings that Bhatnagar wrote as the Committee’s “secretary” indicated, Bhabha was calling the shots in the Committee.¹⁸

What made Bhatnagar defer to Bhabha? Fundamentally, Bhatnagar was a chemist. Even in chemistry, Bhatnagar’s expertise wasn’t in nuclear chemistry, but in colloid chemistry, which found applications in, for example, crude oil drilling. Bhatnagar could not independently evaluate Bhabha’s nuclear energy strategy, and Bhabha gradually took charge of nuclear research funding in India.

Vikram Sarabhai

Though Sarabhai was fairly distinguished, his expertise lay in space research, not nuclear physics.

After Bhabha’s death in 1966, then-Prime Minister Indira Gandhi likely chose him to head the Department of Atomic Energy largely because of his Gujarati heritage, which could give the prime minister political capital against her rival, the Gujarati Morarji Desai.¹⁹

Such a political motive didn’t exist for Nehru, because Desai was not close to Nehru’s level of influence. In any case, ten years Bhabha’s junior and an expert in a different field altogether, Sarabhai fundamentally lacked a good case for heading the Indian nuclear establishment.

Lessons for AI governance

The negative case studies, especially of Bhatnagar, are stark examples of the importance of scientific preeminence—it can substitute for even a massive gap in traditional political power. If an AI researcher possesses great scientific eminence, they could likely stand out in a process to select heads of national technology development programs, even compared to esteemed scientist-bureaucrats whose field of expertise doesn’t happen to be AI. For example, as AI capabilities developments figure more prominently in the public imagination, prominent AI experts like Yann LeCun, Ian Goodfellow or even Andrew Ng could become the US President’s Science Advisor, head the Office of Science and Technology Policy or the National Institute of Standards and Technology, or serve in the White House National Security Council.

As Bhabha’s influence in the Parliament shows, AI experts’ scientific preeminence could also serve as a halo protecting them from criticisms from politicians, who might feel out of their depth to challenge the researcher’s expertise.

Political entrepreneurship and savvy

The archetypal scientist hunkers over her research and dislikes university politics. Bhabha was the rare scientist who showed political entrepreneurship: that is, the willingness to engage in field-building, and political/bureaucratic decision-making.

For example, not content with a research position at the IISc, he applied for a grant to start an institute that had a conscious strategic goal of producing nuclear energy. The Tata Institute for Fundamental Research was crucial to kickstarting nuclear science in India.

He also busied himself with bureaucratic processes, first at the Atomic Energy Committee, then at the Atomic Energy Commission, and finally at the IAEA debating safeguards. At the same time, Bhabha showcased a distinct savvy in his political dealings. Hearing Bhabha’s argumentation at the IAEA, the US Atomic Energy Commission head Glenn Seaborg said that he “was not easy to argue with. Polite but very sure of himself, he was never at a loss for words, and was most articulate. He was a very imposing presence.”²⁰ His securing a highly favorable nuclear deal from the US would have made a “seasoned professional diplomat” proud, according to his subordinate MGK Menon.

In a sense, by the time he was representing India at the IAEA, Bhabha was a seasoned professional diplomat. Because Nehru had given him and Bhatnagar a free hand and a reserve of “hard currency” in international dealings to secure designs, fuels and equipment, Bhabha had training in trading goods that didn’t have publicized prices (being bought and sold, as they were, in a cartel).²¹

Further, Bhabha probably intentionally crafted public and parliamentarian opinion to force Shastri into permitting nuclear weapons research. His handling of the 1964 situation showcased an impressive level of political savvy.

Negative case study

CV Raman

CV Raman, a physics Nobel laureate and Bhabha’s former mentor at the IISc, would have been a fine choice for heading the Atomic Energy Commission, as a matter of superior scientific preeminence. As he later showed by founding and growing the Raman Research Institute, he was skilled in institution-building.

However, Raman didn’t show much political entrepreneurship, i.e., willingness to work with the government. He reportedly smashed the Bharat Ratna award (the highest civilian award in India) the Nehru administration presented him²² and refused government funding for the Raman Research Institute because of his dislike of writing regular update reports. Raman also was “disturbed by the spending of large sums of money in the name of science and scientific research, without proper attention to whether the money was being spent usefully and wisely” in the 1950s, and wrote caustic editorials criticizing the government’s policies.²³

Lessons for AI governance

Political entrepreneurship seems an extremely useful trait for an AI expert who wishes to influence national policy. As Raman’s case illustrates, even an established researcher could find herself unable to affect government policy if they are not willing to work with the government on the inside.

The support of a key decision-maker

Nehru’s support was paramount in catapulting Bhabha to power, especially to a position powerful enough to be able to challenge the authority of Prime Minister Shastri. Nehru was able to get the Atomic Energy Act passed despite some backlash in the Parliament, and usually protected Bhabha from people like his Finance Minister who were troubled by the AEC’s large expenses.

How was Bhabha able to acquire support from Nehru?

Strategic alignment

Nehru’s views on civil nuclear energy

Nehru saw India’s lack of technological progress as the cause of its colonization. More specifically, he believed that a country’s energy production or consumption was a strong signal of its economic growth, and atomic energy was the way to rapidly grow India’s energy production.

In a 1954 interview, Nehru also explained why atomic energy, and not traditional fuels like cow dung or hydroelectric energy, were the panacea to India’s energy problem: “Traditional fuels are cheaper where they occur. But it is a very expensive business where they do not occur. … Yet I could take atomic energy to those areas immediately. … Dr Bhabha said at Geneva that within ten, fifteen or twenty years you may have inexhaustible supplies of fissionable material from sea-water for instance – so that you may really get fairly cheap supplies practically without limits.”²⁴

Bhabha’s views on nuclear energy

Bhabha went a step further than Nehru and believed that there existed a direct causal link between a country’s energy output and its economic development (as opposed to the other way around: economic growth increasing the demand for energy, which raised the energy supply).²⁵ Atomic energy offered a quantum leap in energy generation potential, and the “only chance” of raising standards of living in India and Pakistan’s combined populations of over 450 million.²⁶

Nehru’s views on nuclear weapons

Most of Nehru’s speeches certainly show a commitment to never developing nuclear weapons. For example, as late as 1961 (when tensions between China and India were high, India already knew of the existence of a Chinese nuclear weapons program, and there would be a war between the two countries the next year), Nehru was saying: “We are opposed to atomic bombs, hydrogen bombs, and all that breed. That is not an empty statement for us to make because we will be in a position—we have the competence and the equipment—to make them. Yet, we have said we will not go that way at all. If we had tried hard enough we might have made them. We are probably apart from the three big countries, among two or three others which are so advanced in this matter—maybe one or two countries in Europe and one or two in Asia.”²⁷

In fact, Nehru saw the invention of the atomic bomb as something of an existential risk to humanity. A press report on a 1954 address reads: “The Prime Minister [Nehru] said that the human race was going through a critical phase of its existence, chiefly because of scientific development which had given enormous power to human beings. India was playing an important and dynamic part in this period. Ultimately, the part she played depended upon her own strength.”²⁸

However, Nehru’s interest in a national nuclear program probably stemmed at least in part from a desire to make India a military nuclear power.

First, private correspondence: as early as 1947, Bhabha wrote in an internally circulated note that “[t]he probable use of atomic energy in warfare is likely to revolutionize all our concepts of war and defence. For the moment we may leave that out of consideration except that it makes it absolutely essential for us to develop the methods of using atomic energy for both civil and military purpose.” (emphasis mine)

In the same document, Nehru notes that it doesn’t matter much whether India’s conventional forces number 200,000 or 300,000: “[e]ven the latter figure … is totally inadequate to face a major Power”. Rather, an army that “can expand fairly rapidly if needed” should be the goal, and, among other strategic objectives in Nehru’s planned “line of approach”, an “Atomic Energy Commission should be appointed for research work in the proper utilization of atomic energy and other uses.” Nehru concludes that “the probable use of atomic energy in warfare is likely to revolutionize all our concepts of war and defence. For the moment we may leave that out of consideration except that it makes it absolutely essential for us to develop the methods of using atomic energy for both civil and military purpose.”²⁹ (emphasis mine)

In a note to his Defense Minister Baldev Singh, Nehru said that the “future belongs to those who produce atomic energy” and “[d]efence is intimately connected with this.”³⁰

According to the historian Ashok Kapur, who wrote India’s Nuclear Option, when Nehru approved Bhabha’s memo on the deal with Canada for creating the reactor RAPS-I in 1964, he wrote a note in the margin “somewhat as follows”: ‘‘Apart from building power stations and developing electricity there is always a built-in advantage of defence use if the need should arise.”³¹

In general, Nehru had no real incentives to outrightly declare that he wished to manufacture nuclear bombs. Even if Nehru felt the need to publicly mention a nuclear weapons program for security considerations, he could simply point out that India could build an atomic bomb in a timeline of a few years. With such a statement, he would enjoy about the same deterrent benefits of committing to a nuclear weapons program. One could argue he did something like this in the 1961 speech quoted above, when he said that India was “probably apart from the three big countries, among two or three others which are so advanced in” nuclear energy research.

On the other hand, declaring that he wished to build nuclear weapons capability came with significant costs, among them possible economic sanctions on the underdeveloped country by the US, or a refusal to share designs, materials and training necessary for building a nuclear program.

In light of this, we should consider views expressed in internal memos and private correspondence to be closer to Nehru’s actual beliefs and motivations.

Other times, when cornered by questioners, Nehru would sometimes admit that he was at least open to the idea of a nuclear weapons program. For example, in response to a question on whether the Government of India would keep nuclear weapons in its armory in the future, Nehru replied that “I have no doubt India will develop its scientific researchers and hope Indian scientists will use the atomic force for constructive purposes. But if India is threatened, it will inevitably try to defend itself by all means at its disposal.”³² (emphasis mine)

In a 1946 speech in Bombay, Nehru said that “India will develop her scientific researches and I hope Indian scientists will use the atomic force for constructive purposes. But if India is threatened she will inevitably try to defend herself by all means at her disposal.”³³

In the Constituent Assembly debates on atomic energy in 1948, Nehru said in response to a question that: “Of course if we are compelled to use [atomic energy] for other purposes, possibly no pious sentiments of any of us will stop the nation from using it that way.”³⁴

On the balance, the evidence seems to lean in favor of the view that Nehru knew of the military dimension to India’s nuclear program, particularly the production of plutonium, from the beginning. He welcomed the option to become a “nuclear weapons power” in the future with the technological base that Bhabha was developing.

Bhabha’s views on nuclear weapons

On the balance, it seems that Bhabha aimed for India to develop nuclear weapons capability as early as 1950. For example, Bhabha told the scientist Raja Ramanna, who was hired in 1949, from the outset that “[w]e must have the capability. We should first prove ourselves and then talk of Gandhi, non-violence and a world without nuclear weapons.”³⁵

When PK Iyengar, another former AEC chairman, was asked if Bhabha had been “keen” on India acquiring nuclear weapons, he had said: “Dr Bhabha had in his mind from the very beginning that India should become a Nuclear Weapons State. His emphasis on self-reliance is essentially due to the fact he wanted India to be a nuclear weapons country.”³⁶

In his memorial address, John Cockcroft said that although “it was a declared policy of the government of India not to develop nuclear weapons, and Homi Bhabha of course in his official pronouncements followed the policy of his government,” Cockcroft “always thought, from private discussions, that his attitude was somewhat ambivalent. After the Chinese nuclear bomb test, he certainly wished to put India into the position of being able to make plutonium bombs, if the government so desired.” The new TIFR director MGK Menon, in his memorial address, pointed out that the decision to build the plutonium reprocessing Phoenix plant was taken before the 1962 war. So, the decision could not be security-related. However, as Menon himself conceded, relations between India and China have been testy since the early 1950s. More importantly, India knew of China’s WMD program since before the war.³⁷

There were some choices that Bhabha took with the nuclear program that can be explained better from the hypothesis that Bhabha wanted to develop atomic weapons capability, including the fact that he hired plutonium metallurgists who would be vital to constructing the core of a fission bomb.

Aligned views

As we can see, both Bhabha and Nehru strongly agreed on the civilian benefits of a nuclear program. This was far from a commonly-held position in the recently-deconolonized, still relatively poor country. They also probably agreed that it’d have useful military applications. This was good for Bhabha: Nehru would be much more likely to lobby for Bhabha in the Parliament and in the bureaucracy if Bhabha and Nehru agreed on nuclear strategy.

Shared cultural background

Both Nehru and Bhabha were Cambridge-educated men, bachelors, and shared a love for Western classical music and art.

For example, Bhabha had pushed for Vienna to be the location of the IAEA partly so that he could attend the state opera on IAEA meetings.³⁸ In his college days, Bhabha had almost become a painter. He was still a patron of the arts, slowly building a vast art collection at TIFR.

This shared background contributed to the friendship between the two men. Bhabha in his correspondence addressed Nehru as “My Dear Bhai” (meaning “my dear brother”), while Nehru addressed Bhabha as “my dear Homi”. Indira Gandhi said that Bhabha brought Nehru “warm moments of sensitivity that other people take for granted in their everyday life” that are rare in the outwardly glamorous life of politicians.³⁹

This friendship helped build trust between the two men, and should have helped Bhabha in his ascent to political power.

Negative case study

Meghnad Saha

Meghnad Saha was one of India’s leading nuclear physicists, even having a “bosslike role in science in Calcutta, a city where Congress and Nehru needed to retain an influence amongst their leftist-socialist-communist critics” according to Anderson.⁴⁰

Saha also showcased great political entrepreneurship. For example, he stood for the Lok Sabha elections, and on finding that he was up against a much better funded Congress candidate, wrote to his publisher asking for a Rs 5,000 advance on his textbook Treatise on Heat, “because I am standing for election in the house of the people from NW Calcutta”. Saha won the contest by a margin of 16%.

He had said: “Scientists are often accused of living in the ‘Ivory Tower’ and not troubling their mind with realities and apart from my association with political movements in my juvenile years, I had lived in ivory tower up to 1930. But science and technology are as important for administration now-a-days as law and order. I have gradually glided into politics because I wanted to be of some use to the country in my own humble way.”

To some extent, Saha also had strategic alignment with Nehru. Both believed in the economic potential of nuclear energy.⁴¹

However, Saha specifically disagreed that the AEC should be formed. As he wrote in his biography, India first needed to “grow an independent industrial strength” and “to train personnel in nuclear physics in universities”. To him, the fact that the AEC’s proponents (Bhatnagar and Bhabha) wouldn’t hold research and mentoring jobs in universities was bad.

Even when Nehru asked Saha to join the Atomic Energy Commission, going so far as to phone him to try to change his mind, Saha declined Nehru’s request. So, SK Krishnan became the Commission’s third member. According to Anderson, “Krishnan appears to have played a more passive role in the AEC than Bhatnagar or Bhabha, and certainly a more passive role than Saha would ever have done.” Anderson also points out that though “for generations afterward people thought Saha had been snubbed and excluded, but in fact he had been courted by none other than the prime minister”.⁴²

Even though Saha didn’t join the AEC, he persistently criticized India’s nuclear energy plans in the Parliament. He attacked Bhabha for relying on thorium; he would have preferred a focus on prospecting for uranium. This was a consistent critique till 1952, when he expressed it on the Lok Sabha floor and in his journal Science and Culture.⁴³

Relevance to AI governance

As the Saha case illustrates, it may be important to agree not just with the overarching strategic goal (in this case: “develop nuclear energy”) but also the subgoals (“centralize control in a singular Commission”).

It seems unlikely that Bhabha could have gained the political authority to be able to stand up against Shastri, or to bring the fairly quixotic three-stage plan into action, had Nehru not shielded him against political opposition. Bhabha was able to accumulate the political power to do these things because he agreed with Nehru, not just about the need to build atomic energy, but also about the right way to build it.

In the past, the Chinese president Xi Jinping has stressed on the need to modernize and “intelligentize” the People’s Liberation Army,⁴⁴ while artificial general intelligence found a mention in the 14th National Committee of the Chinese People’s Political Consultative Conference.⁴⁵ The Russian president Vladimir Putin, too, has talked about the importance of controlling the development of AI, saying that “the one who becomes the leader in this sphere will be the ruler of the world.”⁴⁶

If a preeminent AI researcher with political entrepreneurship and savvy agrees with these relatively autocratic leaders, they could quickly bypass traditional, democratic political processes to rise to a position of power. This rapid rise to power might be more difficult to pull off in a country like the US, where distinct groups that may have differing opinions about strategic objectives and approaches, and the suitability of the specific researcher to lead a national project, may be on the same order of political influence. Decision-making power in the Chinese and Russian decision-making apparatuses seem more concentrated in Xi and Putin respectively by comparison.

Foreign connections

While his foreign network was likely not a reason that Bhabha was selected for the job of heading the Atomic Energy Commission, his foreign connections are a key reason he was able to deliver tangible progress.

For example, he was able to get uranium supplies from the UK to fuel the APSARA reactor at highly favorable terms in part due to his friendship with the UK AEC chairman Sir John Cockcroft, whom he knew from his Cambridge days. For the second reactor India produced, in collaboration with Canada, Bhabha’s personal friendship with Canada AEC head WB Lewis again proved useful.

Anderson touches on the importance of having access to informal networks to get any kind of institution-building in nuclear physics done: “India was excluded from information shared by all the nuclear powers, just as the three allies, Canada, Great Britain, and the United States, were keeping things hidden from each other though they were compelled by their agreement to share. (This is why informal networks are so important in nuclear history.)”⁴⁷

Relevance to AI governance

I don’t think a foreign network could be extremely important to an ambitious AI researcher who wishes to influence national policy. This is because in AI, only recently has secrecy become the norm rather than the exception. (For an example of this recent trend, consider that nearly no details about the architecture of OpenAI’s state-of-the-art language model GPT-4 are known.)⁴⁸

It’s important to note, however, that this stands in stark contrast with the history of software development and science research, where open-sourcing and publication is the norm. Even today, almost all industry and academic AI research remains open-source,⁴⁹ and security precautions lax (a case in point is the recent leak of the Facebook LLaMa model weights).⁵⁰

So, as far as knowledge goes, an ambitious AI expert may already have everything they need, given still quite prevalent norms about openness in the AI community.

As far as resources go, a key binding constraint to scaling an AI program seems to be compute.^{51 52 53} This does not seem to be a constraint that having access to foreign scientists will help solve. Compute is not uranium, in that embargoes on compute are publicly known, and subject to legislation where often hundreds of democratic representatives have a stake. It is unlikely (though not completely implausible) that an AI expert could significantly help undo the CHIPS Act,⁵⁴ for example. In contrast, it was much easier for Bhabha to arrange for uranium flows through his personal network.

Works cited

Abraham, I. (1998). The Making of an Indian atomic bomb : science, secrecy and the postcolonial state. London: Zed Books.

Anderson, R.S. (2010). Nucleus and nation : scientists, international networks, and power in India. Chicago: University of Chicago Press.

Associated Press. (2021). Putin: Leader in artificial intelligence will rule world. [online] Available at: https://apnews.com/article/technology-russia-business-artificial-intelligence-international-news-bb5628f2a7424a10b3e38b07f4eb90d4 [Accessed 15 Apr. 2023].

Bhabha, H.J. (1948). Note on the Organization of Atomic Research in India. [online] Indian Defense Studies and Analyses. Available at: https://www.idsa.in/npihp/documents/IDSA-HBP-26041948.pdf.

Bhabha, H.J. and Bohr, N. (1955). The Peaceful Uses of Atomic Energy. Bulletin of the Atomic Scientists, 11(8), pp.280–284. doi:https://doi.org/10.1080/00963402.1955.11453642.

Chengappa, R. (2001). Weapons of Peace: The Secret Story of India’s Quest to be a Nuclear Power. New Delhi: HarperCollins India.

Chowdhury, I. and Dasgupta, A. (2010). A Masterful Spirit: Homi J. Bhabha, 1909-1966.

Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., Casas, D. de L., Hendricks, L.A., Welbl, J., Clark, A., Hennigan, T., Noland, E., Millican, K., Driessche, G. van den, Damoc, B., Guy, A., Osindero, S., Simonyan, K., Elsen, E. and Rae, J.W. (2022). Training Compute-Optimal Large Language Models. arXiv:2203.15556 [cs]. [online] Available at: https://arxiv.org/abs/2203.15556.

Kaplan, J., McCandlish, S., Henighan, T., Brown, T.B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J. and Amodei, D. (2020). Scaling Laws for Neural Language Models. arXiv:2001.08361 [cs, stat]. [online] Available at: https://arxiv.org/abs/2001.08361.

Nuclear Weapon Archive. (2023). Dr. Homi Jehangir Bhabha. [online] Available at: https://nuclearweaponarchive.org/India/Bhabha.html [Accessed 15 Apr. 2023].

OpenAI (2023). GPT-4 Technical Report. [online] Available at: https://cdn.openai.com/papers/gpt-4.pdf.

Penney, W.G. (1967). Homi Jehangir Bhabha, 1909-1966. Biographical Memoirs of Fellows of the Royal Society, 13. doi:https://doi.org/10.1098/rsbm.1967.0002.

Perkovich, G. (2001). India’s nuclear bomb : the impact on global proliferation. Berkeley: University Of California Press.

Sarkar, J. (2017). Sino-Indian Nuclear Rivalry: Glacially Declassified. [online] The Diplomat. Available at: https://thediplomat.com/2017/06/sino-indian-nuclear-rivalry-glacially-declassified/ [Accessed 15 Apr. 2023].

Sciencenet (2023). 朱松纯委员：像‘两弹一星’一样发展通用人工智能—新闻—科学网. [online] Available at: https://news.sciencenet.cn/htmlnews/2023/3/495294.shtm [Accessed 15 Apr. 2023].

Sengupta, R. (2006). ‘Bhabha wanted India to be a Nuclear Weapons State’. [online] Rediff News. Available at: https://www.rediff.com/news/2006/jan/24bspec1.htm [Accessed 15 Apr. 2023].

Sevilla, J., Heim, L., Ho, A., Besiroglu, T., Hobbhahn, M. and Villalobos, P. (2022). Compute Trends Across Three Eras of Machine Learning. arXiv:2202.05924 [cs]. [online] Available at: https://arxiv.org/abs/2202.05924.

Shepardson, D. (2022). Biden signs order on \$52 billion chips law implementation. Reuters. [online] 25 Aug. Available at: https://www.reuters.com/markets/us/biden-sign-order-52-billion-chips-law-implementation-2022-08-25/.

Shevlane, T. and Dafoe, A. (2020). The Offense-Defense Balance of Scientific Knowledge: Does Publishing AI Research Reduce Misuse? Proceedings of the 2020 AAAI/ACM Conference on AI, Ethics, and Society (AIES ’20). [online] Available at: https://arxiv.org/pdf/2001.00463v2.pdf.

Singh, B. (1988). Jawaharlal Nehru on Science and Society: A Collection of His Writings and Speeches. [online] New Delhi: Nehru Memorial Museum and Library. Available at: https://www.indianculture.gov.in/ebooks/jawaharlal-nehru-science-and-society-collection-his-writings-and-speeches [Accessed 15 Apr. 2023].

Takagi, K. (2022). Xi Jinping’s Vision for Artificial Intelligence in the PLA. [online] The Diplomat. Available at: https://thediplomat.com/2022/11/xi-jinpings-vision-for-artificial-intelligence-in-the-pla/.

Venkataraman, G. (1995). Raman and His Effect. Universities Press.

Vincent, J. (2023). Meta’s powerful AI language model has leaked online — what happens now? [online] The Verge. Available at: https://www.theverge.com/2023/3/8/23629362/meta-ai-language-model-llama-leak-online-misuse.

Virk, H.S. (2022). Book Review: C.V. Raman and the Press: Science Reporting and Image Building: Part III: The Raman Research Institute . Current Science, [online] 122(1). Available at: https://www.currentscience.ac.in/Volumes/122/01/0104.pdf [Accessed 15 Apr. 2023].

Wikipedia. (2023a). Gandhism. [online] Available at: https://en.wikipedia.org/wiki/Gandhism.

Wikipedia. (2023b). Manhattan Project. [online] Available at: https://en.wikipedia.org/wiki/Manhattan\_Project.

Zhang, I. (2023). AI Proposals at ‘Two Sessions’: AGI as ‘Two Bombs, One Satellite’? [online] ChinaTalk. Available at: https://www.chinatalk.media/p/ai-proposals-at-two-sessions-agi [Accessed 15 Apr. 2023].

Notes