India's AI Path: From Back Office to Global Player?

Artificial intelligence, often seen as a revolutionary force, is fundamentally a marketing term that has unlocked vast amounts of capital, with trillions of dollars being pumped into large language models. While AI promises automation, potentially reducing workforces by up to 60%, the question remains: does India have a unique AI path, or is it merely following global trends? India boasts the second-largest AI workforce globally, yet it faces significant challenges in resource allocation and investment compared to Western nations and China.

The Human Element in AI Training

In the small town of Karur in Tamil Nadu, a unique approach to AI training is unfolding. Workers in a textile manufacturing factory are seen wearing cameras on their foreheads, often GoPros or Meta glasses, while performing their tasks. This footage is used to train robots. These workers, earning an additional 10,000 rupees per month for this task, are essentially becoming the "fodder" for AI, their everyday activities captured to teach machines.

This practice raises concerns about the dystopian nature of humans becoming data generators for robots. While it might not seem like a traditional tech job, this "bottom end of the tech job" is crucial for training AI, even if the individuals are not directly producing the technology themselves.

Understanding Egocentric Video Data

ObjectBase, founded in 2019, started as a data annotation company and has since evolved into robotics, collecting various types of data, including egocentric data, gripper data, and teleoperation data. Egocentric data, in particular, captures the day-to-day activities of a normal human being, from household chores like cleaning and cooking to personal grooming. This data is vital for building the foundational blocks for robotics, with companies expecting massive volumes of this data to be generated daily.

One such data contributor is Dharni, a 28-year-old teacher who, after having a baby, found herself unable to continue her profession. She now captures her daily activities, such as folding clothes, grooming, and cleaning, for which she is paid 1,000 rupees for three hours of work. She often extends this to six to seven hours to ensure sufficient data is captured. Dharni notes that while many of her friends and relatives have lost their jobs due to AI-driven layoffs in startup and multinational companies, her current role feels secure as long as the AI models are still being trained.

The Extractive Nature of Data Work

The collection of such video data raises fundamental questions about exploitation. While a fair trade might exist when a price is agreed upon, the question of whether individuals are being exploited for their data lingers. There's a concern that this data work could lead to people being pushed out of good jobs into a world where technology, controlled by a few companies, mediates all economic sectors, with these companies reaping the majority of the value.

In some regions of India, workers might be compelled to wear cameras without additional compensation, and the collected data could potentially be used against them. The creation of large language models has also sparked debates about whose knowledge is being used to build these systems, from which companies are now profiting immensely.

The Crucial Role of Data Annotation

Data annotation is the process of explaining information to a computer so it can learn. For instance, self-driving cars need to identify objects on the street, and data annotators label images to train these algorithms. This industry in India has a history spanning about 15 years.

The concentration of engineering colleges and graduates in smaller towns, often from families of farmers or daily wage laborers, presents an opportunity. Companies like NextWave recognized this potential, seeing a large pool of educated individuals who could be trained for data annotation work. Poorani, an assistant deputy manager, highlights how her company works with top retail firms, providing services like data annotation, video annotation, and reinforcement learning through human feedback to train and validate AI algorithms.

The precision required in data annotation is akin to an assembly line, where each screw must be placed accurately. Similarly, data annotators must ensure absolute accuracy in their work.

Bringing Jobs to India's Smaller Towns

While India's major cities like Delhi, Mumbai, and Bengaluru are growth drivers, creating jobs in these metropolises is not feasible for everyone. The traditional societal norms in India, particularly for women, often limit educational and career opportunities. However, data annotation work offers a chance for women to earn a livelihood without leaving their hometowns, significantly adding value to their lives and challenging conservative views. The saying, "If you educate a boy, you educate the family. If you educate a girl, you educate the entire village," rings true as these women contribute to their families' well-being and their children's education.

For individuals in smaller towns, the aspiration to work in IT and BPO sectors often remains unfulfilled due to limited local recruitment. The strategy of bringing work to where the people are, rather than people traveling to where the work is, has been a key approach.

India's Ambitions and Challenges in AI

India harbors significant ambitions for AI, aiming to build its own large language models and products. However, it currently lacks the resources and has not yet made the concerted effort to lead in these emerging AI categories. The nation is still searching for the fastest or cheapest route to gain a competitive edge.

While some reports place India in third place in the global AI race, the true impact of AI on economies will depend on its widespread adoption. India has the potential to be a significant player in this space.

The IT Services Sector and the AI Revolution

Bengaluru, a city built on the tech industry boom of the early 1990s, is a testament to India's IT prowess. The Indian IT services sector contributes significantly to the country's exports, providing a crucial source of dollar income. The availability of skilled engineers, their quick learning ability, and proficiency in English have been key advantages.

For over 25 years, India has served as the "back office" for multinational giants, built on a promise of delivering more than expected. The core offering of India's tech industry has been to solve all technological problems for clients, from routine HR and accounting tasks to software development. The competitive pricing was also a major factor in its success.

However, AI presents the toughest challenge to India's IT services sector, threatening the dominance it has held for decades. A significant portion of the work currently done by the Indian IT service industry is at risk of being fully displaced rather than merely transformed. The job losses expected in the white-collar sector due to AI are anticipated to be substantial.

AI's Transformation of India's IT Landscape

The current AI moment mirrors the threat that the Indian IT services industry faced during the era of neoliberalism. Following liberalization in the 1990s, international brands and products became accessible in India, changing everyday life. The IT industry, as it was known, has been irrevocably altered, and India cannot afford to be complacent.

The need for software engineers to create streamlining business software is diminishing, as large language models can now perform many of these tasks. The high-volume, repetitive tasks common in the IT services industry, where a margin of error might be acceptable, make it a prime candidate for AI substitution. The sector's dominance is threatened because it was not at the forefront of bringing AI to India, being more focused on its existing profitability. A lack of private sector investment in R&D and a perceived lack of faith in their ability to make such investments have been contributing factors.

Tesco's AI-Powered Back Office

Tesco, a major European retailer, operates its architectural design and a significant portion of its operations from its Bangalore-based Global Capability Centre (GCC). This center manages everything from civil, mechanical, and electrical engineering to monitoring over 100,000 refrigerators across its stores through an IoT platform.

One of the most complex applications of AI at Tesco is personalization. With 30 years of data on 22 million customers, AI helps understand patterns and trends to assist customers in shopping better. An AI tool called the "cost intelligence model" was used to analyze the cost of bread ingredients, helping Tesco reduce costs by identifying which components were inflating or deflating in price.

While India continues to perform work for companies based outside the country, often referred to as the "back office," these functions are increasingly being brought in-house within GCCs. AI systems now enable touchless issue resolution in stores, where problems are sensed, categorized, and the right vendor is dispatched to fix them rapidly.

The narrative around AI often focuses on job losses. However, a different perspective suggests that the real threat lies not in AI itself, but in humanity's inability to learn and adapt.

India's Scramble for a Place in the AI Economy

The word "scramble" best describes India's current approach to AI. While the country aims to develop its own applications and solutions, these efforts are not yet widely visible. However, India's IT skills position it to play a significant and unique role in the AI landscape.

India has the potential to become the "AI factory of the world," with the booming AI industry requiring human input. It's possible that the overall number of tech jobs in India could increase, with the country becoming a hub for AI-trained talent globally.

However, leveraging its large population as a cheap workforce to attract foreign tech companies may not lead to long-term sovereignty or resilience in the AI age. Recent market trends show investors fleeing Indian markets, moving capital to countries producing essential hardware for the AI race, such as Taiwan and South Korea. India currently lacks this hardware manufacturing capability.

AI has become a symbol of how the economy can feel rigged to the average person. While India might become the "use case capital of the world," this should not come at the cost of sovereignty. Positioning India as merely a market for US tech giants risks perpetuating its role as a backroom operation, similar to the BPO era.

India is a young country with a population willing to experiment and adapt quickly. For India to achieve its goal of becoming a developed nation by 2047, adopting and adapting AI to its specific circumstances and ensuring its application across all real sectors of the economy is crucial. The nation possesses a unique asset: talent at an immense scale. The challenge lies in ensuring that this skill translates into valuable jobs, otherwise, its true worth remains unrealized.