ILO Warns Against Overinterpreting AI Job Risk Metrics, Calls for Smarter Use of Exposure Indicators

As artificial intelligence (AI), particularly generative AI (GenAI), rapidly reshapes the global workplace, the International Labour Organization (ILO) has issued a cautionary note to policymakers and analysts: widely used AI "exposure indicators" are useful—but far from definitive—tools for predicting the future of jobs.

In a new research brief, the ILO provides a critical assessment of how these indicators are being used to estimate the impact of AI on employment, urging a more nuanced and evidence-based approach to interpreting their results.

From Automation Fears to Cognitive Job Exposure

Exposure indicators have become a central method for identifying which occupations are most likely to be affected by AI. These tools assess how susceptible job tasks are to automation or transformation based on current technological capabilities.

However, the ILO highlights a major shift in findings over time.

Earlier studies—focused on traditional automation—suggested that low-skilled, routine jobs were most vulnerable. But newer AI capability-based models paint a different picture: higher-skilled, knowledge-intensive roles are now among the most exposed.

These include professions in:

• Business and finance

• Computing and data analysis

• Education and training

• Administrative and professional services

This shift reflects the growing sophistication of AI systems, particularly GenAI, which can perform complex cognitive tasks such as writing, coding, and data interpretation.

Ripple Effects Across the Labour Market

The ILO brief goes further, emphasizing that AI's impact is unlikely to be confined to directly exposed jobs. Because occupations are interconnected through shared skills and career pathways, changes in one sector can trigger broader labour market disruptions.

For example, transformations in high-skilled roles may:

• Alter demand for supporting occupations

• Reshape career progression pathways

• Influence wage structures and job quality across sectors

This interconnectedness suggests that AI's effects could be more widespread—and less predictable—than initial exposure estimates imply.

Key Limitations of AI Exposure Indicators

Despite their growing popularity, the ILO stresses that exposure indicators have significant limitations that must not be overlooked.

Among the प्रमुख concerns:

• Static assumptions: Indicators rely on current job descriptions, failing to capture how roles evolve over time

• No adoption context: They do not account for whether AI adoption is economically viable or operationally feasible

• Subjectivity: Many models depend on expert judgments and assumptions about AI capabilities

• No real-world outcomes: Most importantly, they measure what AI could do—not what it will do

As a result, interpreting these indicators as direct forecasts of job losses or labour market disruption can lead to misleading conclusions.

From Prediction Tools to Early Warning Signals

Rather than treating exposure indicators as predictive models, the ILO recommends using them as early warning systems—tools that highlight where change is possible, not inevitable.

To generate meaningful insights, these indicators should be combined with real-world labour market data, including:

• Employment trends and job creation rates

• Wage dynamics and income distribution

• Worker transitions between occupations

• Sector-specific adoption patterns

Broader factors such as national policies, institutional frameworks, and economic conditions also play a critical role in shaping how AI is adopted and its eventual impact on jobs.

Guiding Policy in the Age of AI

The ILO's analysis comes amid growing global concern about the future of work in an AI-driven economy. Governments and institutions are under increasing pressure to design policies that both harness AI's potential and mitigate its risks.

By clarifying the strengths and limitations of exposure indicators, the ILO aims to support more informed policymaking—ensuring that responses to AI-driven change are grounded in evidence rather than speculation.

Experts say this balanced approach is essential to achieving inclusive and sustainable outcomes, where technological advancement translates into better jobs, improved productivity, and shared economic gains.

As AI continues to evolve, the message from the ILO is clear: understanding its impact on work requires more than metrics—it demands context, caution, and comprehensive analysis.