Rising risks to Indian corporates from higher US tariffs: Fitch Ratings todayheadline

The first two infographics in this blog were created with Artificial Intelligence (AI) technology, resulting in production times of mere seconds. In contrast, creating similar infographics without AI would typically require several hours of work.

This illustrates how AI significantly boosts worker productivity. Nearly every sector showcases examples of AI enhancing labour efficiency. For instance, online educational content can be effortlessly generated with AI by just inputting text.

Additionally, it can assume the roles of editors and content writers, significantly reducing the time required to draft a report. These are just a few examples of how AI can improve the productivity of workers. As AI continues to permeate different job fields, it is poised to transform the labour market soon. With the assistance of AI, workers will be more productive, and as a result, fewer workers may be needed to perform the same job. It is also possible that AI will create new jobs.

Recognising the profound impact AI is having on the labour market, since 2023, the International Labour Organisation (ILO) has been evaluating how Generative Artificial Intelligence (GenAI) – a more advanced AI that is enabled to create new content, rather than analysing existing data – might influence various occupations.[1] The ILO accomplishes this through research that enables it to rate the degree of GenAI integration in various professions based on their level of involvement with GenAI. These scores are derived from surveys and interviews with expert panels and workers from multiple fields, aiming to understand the different tasks performed across occupations and how GenAI influences these tasks.

This blog aligns the codes developed by the ILO (henceforth referred to as GenAI score) with Sri Lanka’s Labour Force Survey 2023 to explore how GenAI may impact the labour market in Sri Lanka. The original GenAI scores were assigned to workers in Poland, but mapped to tasks in the ISCO-08, making the scores usable across different countries. The analysis below is merely indicative of how GenAI can affect the labour market in Sri Lanka.

Degrees of integration of GenAI in various job categories

The GenAI score demonstrates how artificial intelligence can effectively perform and support a wide range of tasks in various jobs, highlighting its adaptable and beneficial role in our work lives. The ILO represents GenAI across four levels, with jobs most exposed to GenAI rated as 4, and those with the least exposure rated as 1 (See Figure 1). In addition to these, jobs that have some potential to be influenced by GenAI without undergoing substantial structural alterations are categorised as occupations with ‘minimal exposure’. Subsequently, roles that are entirely unexposed to AI are classified as ‘not exposed’ roles.

Source: Gmyrek et al., 2025

Note: Created using Napkin AI

Sri Lankan context

This study applies the ILO-developed GenAI scores to the 2023 Labour Force Survey data, categorised by industry and occupation codes, to assess GenAI exposure to jobs in Sri Lanka. The findings of this analysis are presented in the section below.

In Sri Lanka, about 1.83 million workers (22.8% of the employed population) are in jobs that have some exposure to GenAI (See Figure 2). This is a little below the global average of 25%.[2] Of these, more than 187,000 workers are in occupations that face the highest level (Gradient 4) of AI, which is slightly below the global average of 3.3% exposure. Among them, the vast majority of 179,290 workers are clerks or clerical support workers, highlighting their vulnerability to potential job displacement by GenAI in the future. Additionally, nearly 142,000 workers (about 1.77%) are part of the group experiencing increasing exposure to GenAI (i.e., Gradient 3). A significant portion of these includes Professionals (68,719 workers), Technical and Associate Professionals (34,859), and Clerks (37,922).

A relatively larger share of workers who are not exposed to AI are in elementary occupations (1,826,845 workers), skilled agricultural and forestry (1,211,408), and craft and related work (1,104,415). Within these groups, nearly 90% of elementary workers, and almost all craft workers, and skilled agricultural and forestry workers (99.9% and 100% respectively), are not exposed to AI—indicating a high degree of insulation from GenAI’s immediate impact.

Source: Authors’ calculations based on LFS 2023 and occupation exposure categories identified by Gmyrek et al., 2025

Note: Figure created using Napkin AI. The same occupation can appear at different gradient levels depending on the tasks performed by various sub-categories of the occupation. Occupational categorisation is based on the International Standard Classification of Occupations (ISCO).

The actual and potential exposure to AI may differ due to several reasons. One reason for this is that employment that could be exposed to AI might be in settings lacking digital technology because of a lack of infrastructure, digital tools, and human skills. Figure 3 illustrates how a lack of digital literacy alone can limit the potential uses of AI in the workplace. The measures of digital literacy and workplace readiness for GenAI use we have employed are not very precise. A more accurate assessment of workers’ and workplaces’ digital preparedness could lower this figure further. For individuals in positions with lower levels of AI exposure (Minimum Exposure, Gradient 1, and Gradient 2), this disparity is more pronounced, indicating a high potential to improve productivity through the use of GenAI for these workers.

Gaps between potential exposure to AI and actual digital literacy among employees

Source: Authors’ calculations based on LFS 2023 and occupation exposure categories identified by Gmyrek et al., 2025

Note: ‘Digitally literate’ employed persons in this analysis are identified using an LFS indicator that captures whether an individual has used the internet within the past 12 months. While this may not be an ideal measure of digital literacy, it serves as a proxy for an individual’s ability to use the internet—and potentially AI—in the future, given the available data. ‘Digital facilities in office’ is captured using an indicator that identifies whether an employed person has used a computer and the internet at their office in the last 12 months.

Conclusion