The Generalist's Blind Spot
Many current reports on AI's economic effects rely on broad categories like "white-collar" or "blue-collar" jobs. This is like trying to understand a city's traffic problems by only looking at the total number of cars, without distinguishing between motorbikes,
buses, and freight trucks. The reality is far more complex. Research has found that AI's impact is not on whole occupations but on specific tasks within them. A single job title can contain dozens of tasks, some of which may be automated while others are augmented or left untouched. By lumping diverse roles together, we get a blurry picture that is nearly useless for creating effective policy, business strategy, or individual career plans. This oversimplification masks who is truly at risk of displacement and who stands to benefit from AI-powered productivity gains.
When Language is More Than Just Words
The vast majority of data used to train leading AI models is in English. This creates a significant "language gap" that has profound implications for a multilingual nation like India, which has 22 official languages and hundreds of dialects. An AI tool that works seamlessly for an English-speaking user in a major city may be completely ineffective for a Hindi or Marathi speaker in a different region. This isn't just about translation; it's about access, opportunity, and equity. The Indian government's Bhashini initiative, which aims to build AI models for local languages, is a crucial step. However, most global impact studies fail to account for this linguistic divide. They implicitly assume an English-first world, which means their predictions about job creation or displacement could be wildly inaccurate for the majority of India's population who do not use English as their primary language for work. Without technology that understands local vernaculars, millions could be excluded from the digital revolution.
A Job Is Not Just a Job Title
Just as language creates divergence, so does the specific nature of an occupation. Two people with the same job title, like "accountant," might have vastly different day-to-day tasks. One might focus on routine data entry ripe for automation, while another specialises in complex forensic analysis that requires human intuition. Studies that fail to make this distinction are missing the point. Recent research highlights that AI often substitutes for routine cognitive tasks while complementing non-routine, analytical work. Therefore, the impact on these two accountants would be entirely different. Granular analysis is essential. Instead of asking if "programmers" will be replaced, we should ask which specific programming tasks are being automated and which are being augmented. This task-based approach reveals a more nuanced reality where AI often acts as a tool, not a replacement, allowing workers to focus on higher-value activities.
The Indian Context: A Magnifying Glass
For India, these methodological blind spots are particularly dangerous. Over 90% of the country's workforce is in the informal sector, a segment almost entirely ignored by mainstream AI impact studies. These workers, from farmers to small-scale retailers, are the backbone of the economy. Yet, policy decisions based on Western-centric, formal-sector-focused AI research risk exacerbating inequalities. The NITI Aayog has rightly pointed out that AI will not automatically benefit the informal sector without deliberate intervention. If our understanding of AI's impact is based on flawed, overly generalised studies, we risk creating policies that fail to support the vast majority of Indian workers. We might invest in retraining programs for the wrong skills or fail to build social safety nets for the most vulnerable populations. The diversity of India's economy demands a much finer-grained analysis.
