Beyond the Hype: What AI Means for Your Career

The recent fall in Indian IT stocks amid concerns that large language model agents, such as those developed by Anthropic, can automate chunks of work currently outsourced to Indian IT services, led to a plunge in the Nifty IT index last week. After Anthropic unveiled a suite of enterprise-focused AI tools, such as "Claude Cowork" and specialised plugins for legal, sales, finance, and data analysis workflows, global software and IT services stocks sold off. Indian IT majors like Tata Consultancy Services, Infosys, HCL Technologies, Tech Mahindra, and Wipro fell by roughly 4–7% in a single session on February 6.

Investors see Anthropic’s tools as capable of automating repetitive coding, testing, contract review, data‑processing, and routine support tasks that form a core part of Indian IT’s offshore‑services playbook. Because these AI agents can run 24×7 at relatively low marginal cost, the market fears clients may shift low‑end, labour‑intensive projects to them, compressing volumes and margins for Indian vendors. 

The AI-related shock is particularly relevant for India, which is often touted as having the advantage of a young population compared to the rest of the world, with a median age of 28. This demographic dividend could disappear amid the AI revolution and turn into a demographic burden.

This concern is heightened because over the last few years India has been facing problems with jobs and growth in real wage rates. Real wages for salaried jobs in India have stagnated or declined in recent years, with data up to mid-2024 showing they were 1.7% lower than pre-pandemic 2019 levels. While nominal wages have increased, high inflation has eroded purchasing power, leading to negative real wage growth in 2020-21 and an uneven recovery since. 

Indian IT companies are already laying off workers, with over 50,000 employees reportedly at risk of job losses this year. This is bad news, as around 80% of India's 50 million white-collar workers are in the IT sector, which indirectly supports the livelihoods of another 250 million blue-collar workers, such as drivers, security guards, and domestic help, thereby further worsening their economic conditions.

But the fear that AI is going to take over all jobs rests on a flawed premise. The belief that artificial intelligence will eventually replace all human jobs rests on a flawed assumption: that AI is a perfect substitute for human workers. This view presumes that humans add no unique value once a task is assigned to AI. Despite rapid advances in technology, this assumption simply doesn’t hold true. Humans and AI make decisions differently, and these differences create powerful complementarities rather than pure competition.

Humans excel at contextual understanding and intuitive judgment. They can make sense of ambiguous situations, use experience‑based knowledge, and adapt quickly when circumstances change. Much of this “implicit knowledge” comes from hands‑on experience and is difficult to codify. However, human decision‑making is not without limitations—biases, fatigue, emotions, and social pressure often cloud human judgment.

AI, on the other hand, thrives in environments rich with data. By processing vast amounts of information, AI can detect subtle patterns that humans might miss. It is highly consistent and does not suffer from fatigue or cognitive overload. Because it scales at low cost, AI can perform millions of decisions rapidly. Yet AI also has weaknesses: it requires large, high‑quality datasets to function effectively, and it struggles in unfamiliar or novel situations where historical data is limited.

A recent study highlights two key types of human–AI complementarity: within‑task and between‑task. Within‑task complementarity occurs when humans and AI work together to produce better outcomes than either could alone. In such cases, AI supports rather than replaces humans—this is augmentation. Between‑task complementarity arises when humans and AI excel at different tasks. For instance, if AI outperforms humans at Task A and humans outperform AI at Task B, the introduction of AI improves efficiency by allocating each task to the party best suited to perform it. This creates a blend of automation and reallocation.

These insights carry important implications for the future of work. Tasks that are repetitive or routine—such as document summarisation or data extraction—are likely to be automated or significantly augmented by AI. But tasks requiring nuanced interpretation, strategic thinking, or context‑sensitive judgment are unlikely to benefit much from automation. Humans will continue to excel in these areas, provided these complementarities remain intact.

However, the persistence of these complementarities depends on how both AI and humans evolve. If AI becomes so advanced at a particular task that humans no longer contribute meaningfully, the task might eventually be fully automated across industries. There is also a more subtle risk: as humans rely more heavily on AI, they may gradually stop engaging deeply with tasks. This can erode the very implicit knowledge that gives them an edge over machines, potentially accelerating their replacement.

For this reason, some experts argue that even when full automation is technically possible, it may not be desirable. Fully automating foundational tasks could limit opportunities for workers to build the skills needed for more advanced roles—areas where human judgment still matters. Over‑automation could inadvertently create a workforce ill‑equipped to handle high‑level tasks where AI alone is insufficient.

In summary, the future of work is unlikely to be a story of AI replacing humans wholesale. Instead, we can expect routine tasks to be automated, moderately complex tasks to involve close human–AI collaboration, and the most complex tasks to remain primarily human‑driven. Rather than fearing an AI takeover, we should focus on understanding how humans and machines can complement each other—and on preserving the uniquely human skills that make that complementarity possible.

There is a need for government intervention by combining reskilling, sector‑specific support, and pro‑employment policies, rather than trying to stop automation. To protect livelihoods, the government must scale up reskilling through programs such as Skill India, offering AI-aligned courses in data science and AI operations. Investment is also necessary to build AI infrastructure, data centres, and semiconductors, and to support startups that can absorb displaced talent into new roles. Rather than resisting AI, India should upgrade its workforce alongside it, not allow it to replace workers.

*Pranjal Chandrakar of Mahindra University co-authored this article