Introduction and background
Welcome to your launchpad for success—where innovation meets action. Disrupt the status quo. Students, businesses, entrepreneurs, and government practitioners unlock growth with applied AI and Robotics, business strategy, and innovation. Join the movement.
Join Nitin Pradhan, visionary founder and former award winning federal CIO and the ex-Managing Director of Center for Innovative Technology, as he delivers bold insights, powerful trends, and untapped opportunities to ignite your business and career based on trusted sources.
Fuel Your Growth. Shape Your Future.
Subscribe now on Substack, Apple Podcasts, Spotify, and more for expert interviews, actionable advice, and forward-thinking strategies that drive real results.
Go Beyond Listening—Start Scaling!
Enroll in our exclusive Udemy courses for deeper learning and mentorship. Want to collaborate or partner? Connect with Nitin on LinkedIn and be part of a thriving growth community.
Let’s get scaling—tune in, take action, and transform your future with ScaleUP USA!
Generative AI and Jobs - A Refined Global Index of Occupational Exposure (ILO Working Paper)
Reference Source: "Generative AI and Jobs: A Refined Global Index of Occupational Exposure" by Janine Berg, Karol Kamiński, Filip Konopczyński, Agnieszka Ładna, Konrad Rosłaniec, and Marek Troszyński (ILO Working Paper).
Key Themes:
Refining GenAI Exposure Assessment: The primary goal is to improve upon previous methodologies for assessing the potential impact of GenAI on jobs by using a more granular dataset of occupational tasks and a mixed-methods approach.
Leveraging Detailed National Data: The study utilizes Poland's 6-digit occupational classification system, which contains significantly more detailed task information (nearly 30,000 tasks) compared to the global ISCO-08 system, to provide a more fine-grained analysis.
Combining Human and AI Expertise: The methodology integrates human assessment (surveys of employed individuals and expert validation) with AI prediction (using Large Language Models - LLMs) to create a robust and accurate knowledge base of task automation potential.
Developing a Global Index from Detailed Data: The study aims to translate the insights gained from the detailed Polish data back to the international ISCO-08 framework to provide updated global, regional, and income-based estimates of employment exposure to GenAI.
Open Methodology for Future Research: The authors intend to make their methodology, including survey instruments and AI model blueprints, publicly available to encourage further research and national-level studies on GenAI's impact.
Most Important Ideas and Facts:
The Study Updates and Refines the ILO's 2023 Global Index: "This study updates the ILO’s 2023 Global Index of Occupational Exposure to Generative AI (GenAI), incorporating recent advances in the technology and increasing user familiarity with GenAI tools."
Utilizes Poland's 6-Digit Occupational Classification System as a Foundation: The analysis begins with Poland's system, which "includes nearly 30,000 tasks. This expands tenfold the number of tasks given in the ISCO-08 structure and enables a fine-grained assessment of tasks’ automation potential..." Poland is considered representative of Central and Eastern European countries and falls between high-income and emerging economies in terms of income and internet access, making it a suitable context for this initial analysis.
Employs a Mixed-Methods Approach: The methodology involves:
Initial Algorithmic Predictions: Using LLMs (GPT-4, GPT-4o, Gemini Flash 1.5) to assign initial automation potential scores (0-1) to tasks in the Polish system.
Human Assessment (Survey): Surveying 1,640 people employed in Poland across each 1-digit ISCO-08 group to rank the automation potential of a representative sample of 2,861 tasks. "We collect 52,558 data points regarding perceive potential of automation for 2,861 tasks." Respondents rated tasks on a 0-100 scale for automation potential by GenAI.
Expert Validation: A smaller group of national and international experts conducted a detailed review and adjustment of the human-surveyed scores through an iterative Delphi-style process.
AI Prediction based on Human/Expert Scores: Inputting the adjusted human and expert scores into an AI model to predict automation potential scores for all nearly 30,000 tasks in the Polish system. This AI assistant was then used to generate scores for tasks in the ISCO-08 system, adjusting the 2023 ILO index.
Focus on Technological Feasibility: The study's assessment of task automation potential is focused on "technological feasibility rather than specific country context," allowing for broader applicability.
Creation of a Verified Knowledge Repository: The iterative expert review process led to the creation of a "knowledge repository" of task automation scores, which was used to train and refine the AI prediction model.
High Correlation Between Predictions: The predictions based on the detailed Polish 6-digit tasks and the ISCO-08 tasks show a high correlation (0.92), indicating the stability of the predictive model and the approach.
Development of GenAI Exposure Gradients: The study introduces a more detailed categorization of occupations into four "exposure gradients" based on the mean and standard deviation of task scores, providing a more nuanced understanding of potential impact:
Gradient 4 (Highest exposure, low task variability): μ ≥ 0.6 and μ - σ >= 0.5
Gradient 3 (Significant exposure, high task variability): 0.5 ≤ μ < 0.6 and μ + σ ≥ 0.5
Gradient 2 (Moderate exposure, high task variability): 0.4 ≤ μ < 0.5 and μ + σ ≥ 0.5
Gradient 1 (Low exposure, high task variability): μ < 0.4 and μ + σ ≥ 0.5
Specific Occupations Identified in Higher Gradients: The Annex provides a detailed list of ISCO-08 occupations by exposure gradient. Occupations with the highest exposure (Gradient 4) include Data Entry Clerks, Typists and Word Processing Operators, Accounting and Bookkeeping Clerks, Statistical, Finance and Insurance Clerks, Securities and Finance Dealers and Brokers, and Financial Analysts. Gradient 3 includes Translators, Interpreters and Other Linguists, Secretaries, Bank Tellers, and various ICT professionals.
Higher Exposure Correlates with GenAI Familiarity and Belief in Impact: Exploratory analysis suggests that respondents who were more familiar with GenAI and believed it would have a significant impact on their occupation tended to assign higher automation scores to tasks.
Limitations and Future Research: The study acknowledges limitations, including the subjectivity inherent in any task scoring system and the need for stronger theoretical assumptions to adjust for respondent beliefs or usage of GenAI. Future research will focus on a detailed analysis of Poland's 6-digit occupations and replicating the method in other countries.
The Index is Generic but Provides a Common Global Denominator: The ISCO-08 based index is generic but its strength lies in providing "a common global denominator and a direct link to national labour forces surveys held in the ILO harmonized micro data repository." This allows for comparable global, regional, and income-based employment estimates.
Key Takeaways:
The study presents a significant methodological advancement in assessing GenAI's potential impact on jobs by moving beyond relying solely on general occupational descriptions. By incorporating detailed task data from a representative national system, combining human and AI intelligence, and creating a verifiable knowledge base, the authors provide a more granular and robust estimate of automation potential. The introduction of exposure gradients offers a more nuanced framework for understanding the varying levels of transformation across occupations. While acknowledging the limitations and the dynamic nature of GenAI, the study provides a valuable updated global index and a reproducible methodology for future research. The publicly available task-level scores are intended to empower further analysis and customized assessments of GenAI exposure.
Share this post