India’s AI vision lofty, capability limited

THE AI Impact Summit held in Delhi (Feb 16-20) is a commendable achievement for India (notwithstanding the organisational glitches on the first day and the Galgotias University fiasco) and burnishes the image of the Modi government in convening summit-level events that relate to global governance in critical domains. Investment commitments of over $250 billion are welcome.

The first such mega AI event to be held in the Global South, it was structured around a unique governance architecture built on three guiding Sutras (principles) — People, Planet and Progress — and seven operational Chakras (pillars) that framed the contours of global cooperation in artificial intelligence.

The preamble to the Delhi Declaration noted that "the advent of AI marks an inflection point in the trajectory of technological evolution. The choices that we make today will shape the AI-enabled world that future generations will inherit."

Endorsed by 88 countries/ international organisations, the declaration called for further international cooperation and multi-stakeholder engagement along the seven Chakras — development of human capital; broadening access for social empowerment; trustworthiness of AI systems; energy efficiency of AI systems; use of AI in science; democratising AI resources; and use of AI for economic growth and social good.

The statement added: "In complementarity with existing international and other initiatives, we will work to foster shared understanding, while respecting national sovereignty, on how AI could be made to serve humanity, noting the initiatives across the seven Chakras (pillars) of the Summit."

AI is a distinctive capability and stands out among technologies, because it is the first general-purpose technology in history that can learn and improve at tasks that previously required human-like understanding, rather than merely executing fixed instructions created by humans. In short, AI has its own 'brain' that can learn on the job, even as it executes various tasks.

In essence, AI is the first technology that can routinely surprise its creators with new capabilities, rapidly absorb almost any cognitive skill when given enough examples and produce open-ended creative output at superhuman volume. The ubiquitous use of ChatGPT and AI-generated text and images (often fake) across the world is but one example of the spread of AI and this is just the beginning.

Further, this AI output is generated while improving quickly, year after year through relatively predictable scaling. No previous technology wave combined those properties at the same time.

This combination makes AI qualitatively different and points to the immense potential for consequential outcomes — both positive and negative; and hence, the global focus on trying to evolve norms, principles, regulations and an ethical framework that would benefit all of mankind. This normative inclusivity was at the heart of the Delhi summit objectives.

The AI domain is dominated by the US and its hi-tech ecosystem that comprises the private sector, academia and government agencies. Thus, it was instructive that despite the lofty aspirational rhetoric in the Delhi Declaration, the US distanced itself from any form of global governance.

The US representative at the summit, White House technology adviser Michael Kratsios reiterated this on February 20, asserting: "As the Trump administration has now said many times: We totally reject global governance of AI." He added that the technology cannot lead to a brighter future if it is controlled by bureaucracies and central authorities.

Regulating new technologies has been central to the distribution of global power and the Cold War is illustrative, where nascent nuclear capability was strictly controlled by the major powers led by the US and the former USSR (now Russia).

The civilian strand of nuclear power was strictly regulated and monitored under UN aegis (the IAEA) and the military domain remained a closed club — of the five nuclear powers: the US, the USSR, the UK, France and China.

The NPT (nuclear non-proliferation treaty) was imposed on all other nations, who were forbidden to acquire nuclear weapon capability and this was an iniquitous treaty, but it maintained a certain degree of stability in the global strategic framework.

Paradoxically, India in the 1960s, despite its limited nuclear capability, with PM Nehru at the helm, was able to contribute to the deliberations on the management of the apocalyptic nuclear weapons. With US support, Delhi was finally accorded a special niche in the global nuclear framework in late 2008.

AI and the nuclear domain are very different, but India has been able to articulate a normative vision of how AI should be harnessed for the larger global good — and this is a lofty, altruistic vision. Global power politics, however, operate in a more contested realist framework and technology denial or strict regulation and limited dissemination is the underlying principle. AI could become the critical determinant in shaping global power and prosperity in this century.

While India seeks to play a significant role in the global AI pursuit, it has a very modest capability and does not rank among the top five AI-capable nations. Yes, India is a major adopter of AI and the crowd surge at the Delhi summit reflects the enthusiasm of the younger Indian demography that comprises entrepreneurs, researchers and startups that are waiting to blossom.

Creating an ecosystem that is enabling and empathetic to nurturing AI in the long term will be the abiding Indian challenge. As Professor Arogyaswami J Paulraj of Stanford University and a global AI leader cautions: "India's heavy reliance on imported information technology — spanning communications, computing and artificial intelligence, from hardware platforms to intellectual property — poses a threat to both its economic health and national security."

Investing in data centres (which need vast amounts of water and electricity) is laudable but this remains at the lower end of the AI ladder. As in the case of acquiring design capability in critical military inventory, what matters in AI is 'compute dominance' and here, the US is at the top of the pack, with China as a distant second. India is invisible. This gap must be bridged if India wants to be an effective AI sherpa for the Global South.