[Tech Talk] AI = Actually Indians

Thought to Be Advanced Automated AI
Actually Manual Work by Indian Engineers
Could Be the Beginning of Human-Machine Collaboration

This is not just a simple pun but a self-deprecating joke gaining popularity in recent IT-related communities. When closely examining the operating principles of cutting-edge automated systems utilizing artificial intelligence (AI), it turns out that Indian operators were remotely controlling them behind the scenes.

There Are 'Indians' Behind the Robots

Office scene in India. Photo is not related to any specific expression in the article. [Image source=Pixabay]

A prime example of AI that was 'actually Indians' rather than machine intelligence is Amazon, the world's largest e-commerce company. Its fully unmanned store, 'Amazon Go,' launched in 2016, became the center of controversy. Amazon Go, with no self-checkout counters or staff, was a new concept store where customers simply picked up the items they wanted and walked out, with payment already completed.

Amazon boasted that it actively utilized thousands of sensors and AI systems to achieve complete unmanned checkout in the store. However, early last month, the IT media outlet 'The Information' exposed the truth about Amazon Go. It was revealed that instead of sensors and AI, 1,000 Indian employees were watching customer videos and manually sorting items one by one.

The 'actually Indians' case is not limited to this. Some companies that claimed to have installed AI systems after massively laying off U.S. headquarters staff to reduce development costs were found to have purchased numerous inexpensive offices in India behind the scenes. There was even a startup that, after failing to develop autonomous robot taxis, came up with a business model of remotely controlling drone taxis from developing countries as a 'reverse strategy.'

India, the World's IT Factory

One of the world's largest IT outsourcing companies, Infosys headquarters in Bengaluru, India. [Image source=Yonhap News]

At this point, one question arises: Why are the personnel replacing robots specifically 'Indians'? This is related to India's industrial model. India is a country where IT-related subcontracting work is considered a key industry. According to Indian trade authority data, as of 2022, India's IT outsourcing service exports reached $150 billion (approximately 205 trillion KRW), accounting for half of the country's total service export volume.

Original source: Ministry of Electronics and Information Technology, India

The countries purchasing IT outsourcing from India are the United States and the United Kingdom. Given their history as former British colonies, India has many personnel proficient in English and engineers familiar with Anglo-American corporate culture. Above all, while the U.S. and the U.K. must pay senior software engineers annual salaries of around $300,000 to $400,000, labor costs in India remain low.

Ultimately, the close 'triangular cooperation' among the U.S. and U.K., which directly develop and export software, and India, which provides highly skilled programmers at affordable prices, forms the foundation of the global IT industry.

Human-Machine Collaboration Is Inevitable... 'Actually Indians' Is Not a Ridicule

Currently, generative AI is too unstable to be deployed in 'professional services' that require uniform quality and perfect, rigorous results. This is mainly due to the well-known phenomenon of 'hallucination'?AI chatbots producing plausible-sounding but actually incorrect information.

When AI hallucinates in reality, it can lead to misjudgments or the spread of false information, so a supervisor or corrector is always necessary alongside AI. Today, outsourcing programmers working hard behind AI to correct its mistakes effectively serve as supervisors.

With current artificial intelligence technology, 'complete automation' is impossible.
[Image source=Yonhap News]

However, the 'actually Indians' label does not necessarily have to be a subject of ridicule. AI is still evolving, and even if AI eventually attains intelligence capable of fully replacing humans, ethical reasons will prevent the disappearance of 'human supervisors.' Just as humans have continued to live alongside machines since the Industrial Revolution, in the AI era, we are more likely to choose collaboration with machine intelligence rather than replacement.

The truly important question is this: Whether fully automated AI or human-machine collaborative tools 'actually involving Indians,' can we enhance our work productivity going forward?

According to Goldman Sachs estimates, AI investment is expected to reach about $200 billion this year, and future AI-related expenditures could account for approximately 2.5 to 4% of the global GDP. If AI proves genuinely useful, advanced countries may seize the opportunity to escape chronic low growth. Conversely, if it does not, the world will have squandered an astronomical amount of money.