AI may automate specific tasks, not entire jobs: IIT Madras research paper

The paper argues that assessing AI’s impact by looking at entire professions can obscure how it changes work and affects different groups of workers.

Artificial intelligence (AI) is more likely to automate specific tasks within jobs than eliminate entire professions, with its impact on employment in India depending on the nature of work, the cost of technology and how organisations restructure roles, according to a new white paper by researchers at the Indian Institute of Technology (IIT) Madras.

The researchers said the paper examines which tasks are being performed or reshaped by AI. (REUTERS/Representative)

The paper, ‘How AI Is Reconfiguring Work in India: A Task-Based Framework for Understanding Augmentation, Displacement, New Roles and the Changing Structure of Jobs,’ made available at IIT Madras website on Monday, argues that assessing AI’s impact by looking at entire professions can obscure how the technology changes work and affects different groups of workers.

The white paper was earlier presented at the ‘Industry Conclave on AI Governance 2026’ organised by the Centre for Responsible AI (CeRAI) of Wadhwani School of Data Science and AI (WSAI), IIT Madras, on the institute’s campus on October 22.

“This paper challenges the idea that AI always takes away jobs. Most jobs have tasks and only some of those can be undertaken by AI. The effect depends on the tasks, the industry, rules for AI and how much it costs to use AI. We also looked at who gains from these changes and who is affected,” said Balaraman Ravindran, head of CeRAI and WSAI at IIT Madras and contributor to the paper.

The researchers said the paper examines which tasks are being performed or reshaped by AI in this particular case, when does that translate into automation or displacement, and how should the job be redesigned because of it? “This paper is intended as the first step in a broader programme of work on AI and employment in India,” the researchers said.

How AI could reshape jobs

Drawing on examples from call centres, manufacturing, small businesses, sanitation, banking, law, software engineering and creative work, the IIT Madras researchers argue that AI’s effects vary across sectors and tasks. They also identify emerging roles such as AI trainers, ethical AI specialists, escalation specialists, AI conversation designers, forward deployed engineers and remote robot operators.

The paper draws on government data, company disclosures, industry vendor reports and journalistic accounts, classifying evidence as measured data, pilot or vendor claims, industry estimates, or forecasts. Examples include a ₹246.72-crore computing allocation to Sarvam, industry estimates of business process outsourcing costs and a Gartner forecast that 80% of customer-service issues could be automated by 2029.

However, the authors acknowledge gaps in the evidence base.

“India has comparatively little rigorous, published labour-market research tracking which specific skill bundles within specific occupations are actually being automated versus augmented,” the paper says, noting that some examples rely on qualitative accounts from industry practitioners rather than quantified benchmarks specific to India.

The researchers caution that the paper’s classification of job-impact pathways should be treated as “a structured hypothesis for future empirical testing, not a settled finding”.

The paper proposes a four-level framework to assess AI’s impact, distinguishing between individual tasks, jobs, occupations and wider labour-market outcomes.

“Most discussion of AI and jobs jumps straight from what AI can do on a single task to sweeping claims about entire professions,” said Gokul S Krishnan, research director at CeRAI and contributor to the paper.

“Our framework separates four levels within the profession - the task, the job, the occupation and the wider labour-market outcome. Seen this way, AI taking over a task does not automatically mean a job disappears. Whether it leads to augmentation, redesign or displacement depends on cost, capability, regulation, safety and how organisations choose to restructure work,” he said.

The paper identifies two broad ways in which people and AI may work together: situations in which AI performs most tasks and humans primarily validate the results, and those in which people remain responsible for decisions while using AI to improve productivity.

Cost of AI could shape who benefits

A key concept in the paper is “AI-cost geography”, which examines how the cost of AI tools and advanced systems influences their adoption across industries and locations.

Affordable AI tools could spread quickly in business process outsourcing, bookkeeping and quality inspection at micro, small and medium enterprises (MSMEs). However, applications requiring advanced AI, such as legal research and medical diagnostic imaging, may remain concentrated in large companies and urban centres with better infrastructure and specialised skills.

The researchers also cautioned that the benefits and risks will be unevenly distributed among entry-level and experienced workers, formal and informal employees, urban and rural communities, and small and large businesses. One particular concern is the effect on entry-level roles that traditionally help workers gain experience and move into higher-skilled positions.

Recommendations for AI adoption in work

The researchers propose three steps to help ensure AI adoption benefits workers and the wider economy.

In the first step, they call for wider adoption of affordable, task-specific AI tools, particularly among MSMEs, to improve productivity and competitiveness. Second, they recommend India-specific research into which tasks and skills AI can automate, while accounting for less visible costs involving quality, trust, safety and accountability. Third, they call for structured pathways to help workers transition into new roles, supported by skills training and greater AI awareness among workers, companies and regulators.

The paper says the challenge is not simply to identify jobs that AI can automate, but to determine how existing roles can be redesigned and how workers can prepare for new responsibilities. It describes the framework as the first step in a broader research programme, with further evidence reviews and field studies involving employers and workers needed to test how widely the identified pathways apply across sectors and occupations.