For India’s AI ambitions, the time to act is now

The good news is that India has much going for it: Engineering skills, a growing digital economy, and tailwinds from the US-China rivalry

On May 17, 2017, AlphaGo, an Artificial Intelligence (AI) system built by Google’s DeepMind, defeated Ke Jie, China’s leading player in the board game, Go. In his book AI Superpowers, Kai-Fu Lee cites this as the seminal moment in China’s AI awakening. Considered the hardest game to master, Go’s dominance by a computer roused the government into action. Within a few months, Beijing announced plans to dominate AI by 2030.

A 2022 study by MacroPolo, a Chicago-based think-tank, finds that India is the third-largest source of AI talent. (Getty Images/iStockphoto)

As origin stories go, perhaps it’s a little too neat. Regardless, six years on, China’s AI leadership is undeniable, and its rivalry with the United States (US) is peerless. Europe is a distant third. India is not even in the race.

And the window for catching up is closing fast. A few months into its release, ChatGPT took over our feeds, spouting off on all matters. It’s not just the know-it-all bot. Software that uses generative AI and large language models is being used across applications, from consumers to enterprises. Computing power, the engine that drives these complex models, has also hit a tipping point. Policymakers in China and the US assumed that AI supremacy would be decided by the end of this decade. At the current pace, the contest may well be over by then.

What about India? A country’s proficiency in AI rests on five factors: Research, money, regulation, data, and talent. India comes up short on most. Tortoise Global’s 2022 AI Index, which ranks countries by AI preparedness, places India outside the top 10 on all measures, except talent. The AI Index Report from Stanford University reaches a similar conclusion — India is great at producing the brains for AI, but average at everything else.

This may appear harsh. Yet our analysis bears this out.

Start with research. By volume, Indian researchers churn out AI papers at a prolific rate — the country is only behind the US and China. But when adjusted for impact and quality, our research is mostly incremental. Next is capital. In 2021, AI companies in the US and China raised $52 billion and $17 billion, respectively. India’s AI startups garnered only $1.3 billion in that same period. And unlike their Chinese or American counterparts, few Indian AI startups were working on transformative problems. Using AI to shorten delivery times or boost sales may lead to frothy valuations, but it does nothing to advance India’s technological sophistication.

India has a clear advantage in terms of brain power. Even restricting to top-tier researchers — those most likely to advance the field — India does well. A 2022 study by MacroPolo, a Chicago-based think-tank, finds that India is the third-largest source of AI talent. Yet, even this good news has a catch — none of this talent stays, with most plying their trade for Big Tech labs in the US. According to our research, the number of highly qualified AI scientists in India is in the low teens. Despite the gloomy assessment, India cannot afford to fall behind in this innovation race. McKinsey, a consulting firm, estimates that countries that are AI leaders will benefit from an additional 20% to 25% in economic growth. Dawdlers may only capture half the upside.

Competence in AI matters for national security also. The foundational nature of this technology makes it applicable across uses: Deep fakes, information warfare, drones, weaponry, and more. Owning the software and hardware pieces of the defence stack will be a competitive military advantage. And since the algorithms are deeply integrated into the systems, our preferred approach of banning technologies from problematic countries is rendered ineffective.

The race for AI supremacy is the race for model supremacy. During the initial phases of machine learning, open-source models such as TensorFlow, and PyTorch encouraged reproducibility and collaboration. This trend is beginning to reverse with large-scale models. The model that underpins ChatGPT is exclusive to OpenAI or Microsoft. Companies are wisening up to the power of these models to transform business and society and want to capitalise on it. Without the capability to develop its homegrown models, India will be more reliant on Big Tech than ever before.

The good news is that India has a lot going for it: Engineering skills, a growing digital economy, and tailwinds from the US-China rivalry. This week, the government announced a task force to draft a road map for the country’s AI ecosystem. It will need to move fast.

Shailesh Chitnis is an ex-entrepreneur and a fellow in high-tech geopolitics at Takshashila Institution. This is the first of a two-part series on India’s AI preparedness.

The views expressed are personal.