Will India import AI Brain as a service, like it buys oil? Should India continue to overlook that it contributes to the training of global AI models without developing its own? Can India afford to miss out on the potential of AI by remaining a passive bystander? These questions demand introspection and action as India stands at a critical juncture in the global AI race.
India must invest in an Indic AI model or a similar home-grown capability if the country wants to stay in the global AI land grab. It's time for India to make a multi-billion-dollar bet on developing its home-grown AI.
A Treasure Trove for Global AI Models
With its 1.4 billion-strong population, diverse linguistic and cultural landscape, and rapid digital adoption, India has become a veritable training ground for the world's AI models. Moreover, the sheer scale and diversity of data generated by Indian users have made the country a fertile ground for global AI giants, from Chinese apps to OpenAI, to test and refine their algorithms.
These AI models, in turn, learn from the vast and varied interactions of India's population, honing their capabilities and becoming ever more sophisticated. This wealth of data is invaluable in training AI algorithms to understand context, nuance, and regional differences, ultimately benefiting the companies that deploy these AI models worldwide.
Turning the Tables: The Case for India's Home-grown AI Models
By harnessing its vast human resources and cultural diversity, India has the potential to build AI models that cater to its specific needs, challenges, and opportunities.
Consider the following advantages of investing in home-grown AI models:
Cultural and Linguistic Sensitivity: With over 1,600 languages spoken across the country, India's linguistic diversity poses a unique challenge for AI models. Developing home-grown AI models can help address this challenge by training algorithms to understand and process the nuances of India's various languages, dialects, and cultural contexts.
Customised Solutions: India's diverse population and varying regional needs call for AI solutions tailored to the specific challenges faced by different communities. Home-grown AI models can be designed to cater to these unique needs, providing solutions that are better suited to the Indian context.
Data Sovereignty: As the nation's data becomes an increasingly valuable resource, India must maintain control over its own data and ensure that it is used to benefit its citizens. By developing home-grown AI models, India can assert its data sovereignty and ensure that the nation's data is used to drive local innovation and address pressing challenges.
Economic Benefits: Investing in home-grown AI models can have significant economic benefits for India, including job creation, fostering innovation, and boosting global competitiveness. By nurturing its AI ecosystem, India can attract international investments, develop a thriving AI industry, and position itself as a global AI leader.
How Indians Trained a Generation of China's AI Models
As an AI enthusiast, I've been following the rise of China's app factories for years. My colleague Shadma Shaikh at FactorDaily has been flagging the insidious rise of Chinese apps over the past decade, much before they were banned by the Indian government.
What started as a seemingly innocuous trend transformed into a tidal wave, with Chinese tech giants like ByteDance and Tencent churning out apps such as TikTok and Kwai, which have taken the world by storm. But there's a side to this story that has yet to be explored: the role that India, and its massive population, have played in the rise of these AI-driven platforms.
With its 1.4 billion people, numerous languages, and diverse cultural landscape, India has provided an unparalleled opportunity for these Chinese app factories to refine their algorithms and train their AI models. Millions of Indian users have unwittingly contributed to the development of these apps, their data feeding the machine-learning models that power the platforms' content recommendation engines, language processing capabilities, and more.
I've seen first-hand how the likes of TikTok and Kwai have captivated the Indian audience, their addictive algorithms keeping users hooked and coming back for more. But with each video watched, liked, and shared, Indian users have inadvertently played a role in the rise of China's AI empire. The data generated by their usage has been instrumental in fine-tuning these platforms' AI models, ensuring they become more effective and efficient with each passing day.
Where the World's AI Comes to Train
How many users does ChatGPT have? And how many of them are based in India?
Depending on who you ask, the answers can range from 10 million to more than 50 million Indian users. It's not just China's app factories that have recognised the potential of India's vast and diverse population. New-age AI models, such as those developed by OpenAI, have also tapped into the Indian market, gaining millions of users who, knowingly or unknowingly, contribute to the ongoing improvement of these algorithms. This is both fascinating and somewhat disconcerting, as the potential for India to develop its own home-grown AI models has yet to be explored.
I can't help but wonder: what if we harnessed this immense potential to develop our own AI models tailored to our unique needs and challenges?
Imagine the possibilities of AI algorithms that understand the nuances of our various languages, dialects, and cultural contexts, offering solutions specifically designed for the Indian market. Rather than fuelling the growth of foreign AI platforms, our data could be used to drive local innovation and address pressing challenges.
The rise of China's AI-driven platforms, fuelled partly by India's vast population and diverse data, should serve as a wake-up call for our nation. Suppose we do not act now to invest in our own AI infrastructure. In that case, we risk being left behind in the global AI race, our valuable data continuing to benefit foreign tech giants, while our AI ecosystem still needs to be developed.
AI: The New Battlefield for Talent and Innovation
India is the world's third-largest military spender, with a defence budget of $67.1 billion in 2020, accounting for 3.7% of global defence spending. The country's strategic position, geopolitical challenges, and aspirations for regional influence drive its substantial defence investments. However, as AI emerges as the next frontier in modern warfare, India must shift its focus towards building a home-grown AI infrastructure to ensure its place on the global stage.
The future of warfare will be shaped and influenced by AI, as nations compete in a high-stakes race to develop cutting-edge technologies and harness the power of data-driven insights. This battle will be fought not just with traditional military might but also with innovative research, talent development, and AI-driven capabilities.
AI has the potential to transform every aspect of the defence sector, from intelligence gathering and surveillance to decision-making and combat operations. Countries that can leverage AI in their defence strategies will gain a significant advantage over their adversaries, making the race for AI supremacy a critical battle to win.
A Loss of Identity: The Cultural and Linguistic Cost of AI Dependency
India's linguistic diversity is nothing short of astounding. The country is home to over 1,600 languages, with 122 spoken by more than 10,000 people each. This rich tapestry of languages is a testament to India's vibrant cultural heritage and its long history of regional variation.
However, this vast linguistic landscape also presents a unique challenge for AI models, particularly those designed and developed outside India.
An AI model that doesn't consider India's language diversity risks leaving a significant portion of the population behind. For instance, while Hindi is the most widely spoken language in India, it is spoken by only 41% of the population. Similarly, English, though commonly used for official and administrative purposes, is spoken by only 10% of Indians. This means that relying on AI models that cater predominantly to Hindi and English speakers would exclude many people from benefiting from AI-driven solutions.
Existing AI systems like GPT-4, designed to comprehend and generate text in several languages, often struggle to capture the nuances of India's linguistic and cultural mosaic. By training AI systems on the country's vast array of dialects, idiomatic expressions, and cultural references, India can ensure a more inclusive and accurate representation of its people.
Moreover, if we do not invest in AI models that understand and cater to India's linguistic diversity, we risk losing a wealth of social and cultural history. Language is essential to a society's values, history, and identity. AI models that cannot appreciate the nuances of India's many languages may inadvertently contribute to the erosion of this rich heritage.
By developing AI models that can process and understand the full spectrum of India's languages, we can ensure that these technologies serve as a bridge, rather than a barrier, to India's linguistic heritage.
Final words of caution come from Umakant Soni, the general partner of the $100 million ART Venture fund that backs early-stage AI and robotics companies. "Given the lead in the knowledge economy, we have a great foundation with the knowledge to power AI and eventually the upcoming 'Intelligence Economy', which will create $15.7 trillion of new economic value. However, unless we act quickly and decisively, we might only train the AI brain built by ChatGPT/OpenAI with human feedback. In the future, we will not only pay for oil, but also for using this AI brain."
(Pankaj Mishra has been a journalist for over two decades and is the co-founder of FactorDaily.)
Disclaimer: These are the personal opinions of the author.
