The International Labour Organization (ILO) research brief Workers’ exposure to AI: What indicators tell us – and what they don’t1 analyses "AI Exposure Indicators" to estimate how artificial intelligence may substitute or transform specific human tasks. It highlights a critical shift: while earlier automation primarily affected routine manual labor, current AI capabilities disproportionately expose high-skill, high-wage cognitive occupations in finance, computing, legal, and managerial sectors.
The brief warns that these highly exposed roles occupy central positions in occupational networks, meaning AI-driven changes in one "hub" (e.g., analytical or administrative) can trigger indirect spillover effects across entire career paths. Governments are urged to use these measures as "early warning indicators" for task transformation rather than direct predictors of job loss.
Key Pillars of the AI Exposure Framework
Cognitive-Analytical Exposure: Identifying business, mathematics, and education as the most exposed sectors due to AI's proficiency in "brain work".
Indirect Network Effects: Mapping how exposure in central professional fields impacts related roles through shared task requirements and occupational transitions.
Methodological Triangulation: Combining expert-based assessments, patent-mapping, and GenAI self-assessments to create a "Mean Normalised Measure" of exposure.
Static vs. Dynamic Task Lists: Acknowledging that current indicators rely on static task descriptions and may not capture future workflow reorganizations.
Economic Feasibility Gap: Distinguishing between what is "technically possible" to automate and what is "economically viable" to adopt based on cost-benefit dynamics.
Inclusive Transition Policy: Using exposure data to design proactive skilling and social protection strategies that mitigate rising inequalities.
What is the "Mean Normalised Measure"? The Mean Normalised Measure is a statistical aggregate that combines different AI exposure methodologies (expert, patent, and AI-based) to provide a more stable estimate of task vulnerability. It provides the "Technical Fidelity" needed to see that exposure is not a binary risk; rather, it identifies a high-exposure "U-shape" where both highly cognitively demanding jobs and certain lower-skill office support roles face significant task transformation. By normalizing these disparate scores, the ILO offers a more grounded baseline for cross-country labor market comparisons.
Policy Relevance: India’s Service-Led Economy & Task Transformation
The "IT-BPM" Exposure Hub: India’s high-wage, cognitive-intensive service sector (IT, Finance, and Analytics) aligns perfectly with the ILO’s "High Exposure" profile. This provides a primary mechanic for the Ministry of Electronics and Information Technology (MeitY) to audit which "Business Process" tasks are most substitutable versus those requiring human-in-the-loop critical thinking.
Bypassing the "U.S.-Centric Data" Trap: The ILO’s warning about U.S.-based data suggests that India must develop a localized Occupational Task Database. Without this, India risks "Implementation Infidelity" by applying skilling interventions to jobs that may have different task compositions in a developing economy context.
Mechanical Link to "Middle-Income" Resilience: Since high-wage cognitive jobs are the most exposed, the brief identifies a potential "Strategic Barrier" to India’s middle-class growth. If entry-level analytical roles are automated, the mechanical "stepping stones" for career progression in sectors like Legal and Finance may be disrupted, requiring a policy shift toward "Task-Based" rather than "Job-Based" training.
Operationalizing "Early Warning" Systems: Using the ILO's Mean Normalised Measure, Indian labor agencies can map "vulnerability clusters" in urban hubs (like Bengaluru or Pune). This acts as a mechanical prerequisite for designing targeted Social Protection for "Office Support" roles that show high exposure variation but low career-path connectivity.
Follow the full brief here: ILO: Workers' exposure to AI - February 2026
