AI’s costs fall unevenly across workers; IIT Madras paper flags surveillance, work stress

A new whitepaper from the Centre for Responsible AI at the Indian Institute of Technology Madras reveals that the costs of artificial intelligence adoption among workers in India are not evenly distributed. Entry-level jobs, in particular, are at risk due to the impact of AI automation, according to the paper titled "How AI Is Reconfiguring Work in India: A Task-Based Framework for Understanding Augmentation, Displacement, New Roles and the Changing Structure of Jobs."

The research highlights differences in how AI affects various types of workers, including entry-level, experienced, formal, informal, urban, and rural workers, as well as those employed in micro, small, and medium enterprises (MSMEs) versus large firms. Large companies are better equipped to absorb the fixed costs of integrating and customizing advanced AI models, while MSMEs rely more on narrow, subsidized tools.

The paper also points out disparities between work conducted in English and regional languages, noting that AI systems perform less effectively on nuanced, code-switched conversations in Indian languages. Beyond task automation, the study identifies new roles emerging due to AI, such as robot operators and AI escalation specialists.

However, it cautions that the evidence does not systematically track whether these roles provide stable employment or are piecework-based, which would impact how productivity gains are distributed among workers.

Job quality issues such as wages, surveillance, work intensity, stress, burnout, career progression, human interaction quality, and error responsibility are areas that the paper acknowledges require further investigation. Specifically, the authors question whether AI will introduce new performance tracking methods, like scoring calls handled by BPO employees after AI takes over routine queries, or if AI will result in workers dealing with a higher volume of exceptions following automated tasks.

The paper raises concerns about the lack of systematic Indian data measuring job quality factors. For junior roles, traditionally serving as apprenticeships that help new employees develop skills for senior positions, the introduction of AI could automate tasks such as password resets, basic contract reviews, and boilerplate coding.

This may raise questions about how individuals entering the workforce will gain the necessary experience. One potential solution could be redesigning junior roles to focus on supervising and correcting AI systems, rather than performing routine tasks, but it remains unclear if firms will make such changes or simply hire fewer junior employees.

The study primarily relies on formal-sector evidence, despite informal work making up a larger share of India's workforce compared to examined formal jobs. AI could indirectly affect informal workers through prices, consumer demand, credit access, platform management, and business survival. For example, job losses in formal IT and BPO sectors could reduce local spending supporting informal service providers.

Additionally, AI-based credit scoring could expand formal credit access for some informal workers while excluding those without digital transaction records. The paper notes that these effects have not been quantified and identifies informal employment as a priority for future research. The authors call for India-specific research into AI's labour-market effects and structured pathways for workers to transition as jobs change.