Policy Research Working Paper 10863
Buffer or Bottleneck?
Employment Exposure to Generative AI
and the Digital Divide in Latin America
Paweł Gmyrek
Hernan Winkler
Santiago Garganta
Poverty and Equity Global Practice A verified reproducibility package for this paper is
July 2024 available at http://reproducibility.worldbank.org,
click here for direct access.
Policy Research Working Paper 10863
Abstract
Empirical evidence on the potential impacts of generative findings show that certain characteristics are consistently
artificial intelligence (GenAI) is mostly focused on high-in- correlated with higher exposure. Specifically, urban-based
come countries. In contrast, little is known about the role of jobs that require higher education, are situated in the formal
this technology on the future economic pathways of devel- sector, and are held by individuals with higher incomes
oping economies. This paper contributes to fill this gap by are more likely to come into interaction with this technol-
estimating the exposure of the Latin American labor market ogy. Moreover, there is a pronounced tilt toward younger
to GenAI. It provides detailed statistics of GenAI exposure workers facing greater exposure, including the risk of job
between and within countries by leveraging a rich set of automation, particularly in the finance, insurance, and
harmonized household and labor force surveys. To account public administration sectors. When adjusting for access
for the slower pace of technology adoption in developing to digital technologies, the findings show that the digital
economies, it adjusts the measures of exposure to GenAI divide is a major barrier to realizing the positive effects of
by using the likelihood of accessing digital technologies at GenAI on jobs in the region. In particular, nearly half of the
work. This is then used to assess the extent to which the positions that could potentially benefit from augmentation
digital divide across and within countries will be a barrier are hampered by lack of use of digital technologies. This
to maximize the productivity gains among occupations negative effect of the digital divide is more pronounced in
that could otherwise be augmented by GenAI tools. The poorer countries.
This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to
provide open access to its research and make a contribution to development policy discussions around the world. Policy
Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at
hwinkler@worldbank.org. A verified reproducibility package for this paper is available at http://reproducibility.worldbank.
org, click here for direct access.
RESEA
CY
LI R
CH
PO
TRANSPARENT
ANALYSIS
S
W
R
R E
O
KI P
NG PA
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Buffer or Bottleneck?
Employment Exposure to Generative AI and the
Digital Divide in Latin America1
Paweł Gmyrek, Senior Researcher, ILO
Hernan Winkler, Senior Economist, World Bank
Santiago Garganta, Senior Researcher, CEDLAS-UNLP
© 2024 The World Bank and the International Labour Organization
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage
the exchange of ideas about development issues. An objective of the series is to get the findings out
quickly, even if the presentations are less than fully polished. The papers carry the names of the
authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in
this paper are entirely those of the authors. They do not necessarily represent the views of the
International Bank for Reconstruction and Development/World Bank and its affiliated organizations,
or those of the Executive Directors of the World Bank or the governments they represent, or those of
the ILO.
1
We would like to thank Richard Samans, Sergei Suarez Dillon Soares, Janine Berg, Leonardo Iacovone, Paulo
Bastos, William Maloney, Carlos Rodriguez Castelan, and staff from ILO Research and from the World Bank
Poverty team for valuable comments. We also thank ILO STATISTICS team for data contribution, particularly
David Bescond, who produced several inputs to our estimates based on ILO micro data repository, as acknowledged
in detail in the paper. This article is a product of the staff of the World Bank and the ILO. It has been released both
in the World Bank Policy Research Working Paper Series, and in the ILO Working Paper Series
(https://doi.org/10.54394/TFZY7681, ISBN: 9789220410035). It also serves as a background paper for the World
Bank Regional Study “Digital Equality: A Pathway for Inclusive Growth in LAC”.
1. Introduction
Public attention to Generative AI (GenAI) has been on the rise since the introduction of the
conversational models, such as ChatGPT, Bard or Gemini. The impressive abilities of the Large
Language Models (LLM), followed by other neural network-based AI systems capable of
generating image and even video from simple text prompts have raised a range of important ethical
and security questions for national policy makers and international cooperation structures.
However, the topic that captures most daily attention of regular citizens is the potential impact of
these quickly advancing tools on jobs.
In the United States (US), over half of all adults are more worried than excited about AI in daily
life, citing the “loss of human jobs” as their most important concern (Faverio and Tyson, 2022;
Pew Research Center, 2023; Rutgers, 2024). In Switzerland, a 2023 survey focused specifically
on GenAI revealed that of 1,000 respondents already working with a computer, almost half (43%)
were concerned about losing their job in the next five years, with those frequently using GenAI at
work being disproportionately (69%) more concerned (Grampp et al., 2023). This suggests a rapid
departure from more positive assessments of AI in surveys collected by OECD prior to the arrival
of publicly accessible chatbots in late 2022 (Lane et al., 2023; OECD, 2023). 2
Not surprisingly, the potential transformation that might result from the interaction of GenAI with
labor markets has also attracted growing attention among scholars. Main research questions have
centered around the impact on employment, emerging occupations, productivity and job quality.3
A recent paper from the IMF provides a comprehensive overview of this literature, at the same
time highlighting the scarcity of studies that go beyond high-income countries (HICs) (Comunale
and Manera, 2024).
Bridging this research gap, our study provides new evidence on the potential impacts of GenAI
across labor markets in the Latin America and the Caribbean (LAC) region. Building on the
approach developed by Gmyrek, Berg and Bescond (2023) – GBB hereafter – we provide new
evidence on AI exposure between and within countries by leveraging harmonized household and
labor force surveys for LAC from the World Bank (WB) and the International Labour Organization
(ILO). By building on the comparative strengths of the datasets from both institutions, we develop
a complete regional overview, accompanied by country-level estimates of the potential
occupational exposure, with further breakdowns by detailed demographic and labor market
characteristics.
An important contribution of this study is to provide a first attempt at adapting measures of jobs’
exposure to GenAI to the context of developing countries, where even workers in occupations that
are generally expected to benefit from GenAI may not be able to reap its benefits due to poor
access to digital infrastructure. We implement this adjustment by estimating measures of computer
2
In OECD’s survey of workers “four in five workers said that AI had improved their performance at work and three
in five said it had increased their enjoyment of work” (…) “Workers were also positive about the impact of AI on their
physical and mental health, as well as its usefulness in decision making” (OECD, 2023).
3
E.g., see Brynjolfsson et al., 2023; Hui et al., 2023, Beraja et al., 2023, Adams-Prassl et al., 2023.
2
use at work across ISCO 2-digit occupations, workers and country-level characteristics based on
PIAAC data and by subsequently imputing them into individual observations in country-level
surveys included in the SEDLAC database. We then use this measure to create two categories
among workers who are expected to benefit from GenAI use because of the nature of their
occupations: those who have access to digital technologies, and those who do not. The size of the
latter is an indicator of the number of workers who will not be able to enjoy the productivity
benefits of GenAI even though their jobs could theoretically benefit from the transformation. We
also discuss the detailed demographics of the groups that are most likely to be negatively affected
by these infrastructure limitations.
Our findings indicate that between 30 and 40 percent of employment in the LAC is exposed in
some way to GenAI. This exposure is linked with the economic status of countries, suggesting that
income levels are a strong correlate of GenAI’s impact on labor markets. This total level of
exposure includes three categories: exposed to automation, augmentation, and “the big unknown”.
The latter includes occupations, which – depending on the progress of technology and the use of
adjacent technological applications, such as LLM-based agents – could fall closer to automation
or augmentation.
Certain characteristics consistently correlate with higher overall GenAI exposure. Specifically,
urban-based jobs that require higher education, are situated in the formal sector, and are held by
individuals with higher relative incomes are more likely to come into interaction with this
technology. The share of jobs exposed to automation is relatively small but nontrivial at about 2
to 5 percent of total employment. Younger and female workers tend to face greater automation
exposure, particularly in the finance, insurance, and public administration sectors. At the same
time, the shares of jobs that could benefit from a productive transformation with GenAI are
consistently higher than those with automation risks across all LAC countries, ranging between 8
and 12 percent of employment across countries. This is particularly the case for the jobs in
education, health and personal services. In addition, the sectors oriented towards customer service
(retail, trade, hotels, restaurants, etc.) face an elevated exposure to "the big unknown". This
category encompasses the largest (14-21 percent) share of employment in our estimates,
demonstrating that, while the concept of occupational exposure is easier to establish, the precise
effects on how many occupations might evolve are harder to predict for a large share of today’s
labor markets.
Finally, we find that access to digital technologies is a critical determinant of the extent to which
workers can harness the potential benefits of GenAI. Nearly half of the positions that could
potentially benefit from augmentation are hampered by digital shortcomings that will prevent them
from realizing that potential. Specifically, 6.24 percent of jobs held by women and 6.22 percent of
those held by men are affected due to these gaps. Similar limitations apply to the jobs in the “big
unknown” category: even though some of them could potentially pivot towards augmentation
through increasing complementarity between GenAI and the human worker in these occupations,
the digital gaps will prevent large shares of these jobs from such a scenario.
The rest of this study is structured as follows: section 2 provides a general overview of the LAC
region and elaborates on the theoretical effects one could expect from the interaction of GenAI
3
with its labor markets, section 3 discusses the data and methods applied to our analysis, section 4
provides a detailed breakdown of our findings, with the final discussion presented in section 5.
2. LAC region and the theoretical effects of GenAI
The definition of the region of Latin America and the Caribbean (LAC) can have a varying scope
across different institutions. In the case of our study, we rely on a heuristic approach of including
the maximum number of countries for which we can find data of sufficient quality in the databases
of the WB, ILO and any other relevant sources. The final sample includes 22 countries, shown in
Figure 1 according to their income-based grouping used by the WB in 2022, and total population.
The region is very heterogeneous, from very small islands in the Caribbean with fewer than half a
million inhabitants, to countries with large populations such as Brazil and Mexico. Accordingly,
it ranges from high-income countries such as Uruguay and Panama to lower-income countries such
as Nicaragua and Honduras.
Figure 1. GDP per capita, population and income status of LAC countries in the sample
While there is a large body of literature analyzing the impacts of technological change on the labor
market outcomes of LAC (for example, see Dutz et al. 2018), the expected incidence of GenAI is
likely to be different from that of previous technological breakthroughs. Autor (2024) claims that
the transformational impact of new technologies on labor is through the reshaping of human
expertise, and he illustrates this hypothesis with two examples: the adoption of mass production in
the 18th and 19th centuries, and the adoption of digital technologies since the 1960s. The
emergence of mass production changed the complex work of artisans into self-contained and
simple tasks carried out by production workers, using new machinery, and overseen by others with
higher levels of education. The increased demand for this “mass expertise” was accompanied by
an increasing number of high-school graduates, leading to the rise of a new middle class. Later,
4
digital technologies allowed to carry out routine tasks by encoding them in deterministic rules.
Non-routine tasks could not be replaced by this technology because they are not attained by
learning rules, but through learning by doing. As a result, digital technologies gave rise to a new
form of expertise by allowing professionals to obtain and process information more efficiently,
and thereby having more time to interpret and apply it. The routine jobs replaced by this technology
tended to be in the middle of the earnings distribution, while the non-routine jobs complemented
by digitalization tended to be at the top, leading to a polarization of the labor market. AI, in
contrast, can perform non-routine tasks that often require tacit knowledge. For example, it can
allow non-elite workers (such as nurses) to engage in complex decision-making, and it can
automate some of the tasks carried out by high-skill workers such as doctors, software engineers
and lawyers. However, as described below, the final impacts on jobs will depend on other factors
as well. For example, the direct automation impacts of GenAI on jobs may be offset by positive
impacts on productivity, which would strengthen labor demand.
While no previous granular assessments of occupational exposure to GenAI exist for the LAC
region, there have been comparisons to other regions made in broader studies. For example, GBB
(2023) place LAC somewhere in the middle of the regional ranking of potential automation
exposure, with 2.5 percent of total employment falling into this category (Figure 2). In terms of
augmentation potential, the same study ranked LAC as the third from the bottom (12.8 percent of
employment). Similarly, while the WEF (2023) global study did not provide a specific regional
ranking, it projected a 5-year structural labor churn in LAC at 22 percent, slightly below the global
average (23 percent). In other words, the LAC region can be characterized as having economies
with an average level of exposure to GenAI that is less than that of the most industrialized nations,
yet higher than that found in low-income regions, making it a relevant intermediate benchmark.
Figure 2. Automation and augmentation potential: LAC vs other regions
Source: Gmyrek et al., 2023
In theory, the rise of GenAI and its potential positive impacts on labor productivity could pose a
significant opportunity for developing countries. Some recent private sector studies even suggest
that aggregate impact of widespread AI adoption could add between 0.1 and 1.5pp of annual
productivity growth in HICs, with slightly lower figures estimated for Emerging Markets (EM)
5
(Goldman Sachs, 2023; McKinsey, 2023). Such projections might be particularly enticing for the
LAC region, which has long grappled with a persistent productivity gap in comparison to other
areas of the world. While the developing nations in Asia and Europe managed to narrow their
productivity gap with the United States between 1990 and 2019, such gap increased for the LAC
region during the same period (IMF, 2022). Recent trends also raise concerns, since, despite some
country variation (Erumban et al., 2024), the overall productivity growth has been almost zero in
LAC ever since the start of the global productivity slowdown of the last 10 years (Dieppe, 2021).
Compared to other regions, barriers to innovation and technology adoption have been particularly
salient factors limiting productivity growth in LAC.
Could GenAI help unlock this productivity impasse? Recent empirical studies focused on the use
of GenAI in particular occupational settings suggest that the positive impacts on productivity can
be large. For example, Peng et al. (2023) implemented a controlled experiment among professional
programmers and found that access to a GenAI assistant reduced the time to complete
programming tasks by 56 percent. Brynjolfsson et al. (2023) find that access to GenAI increases
productivity among customer support workers in terms of issues resolved per hour, which is driven
mostly by the boost of performance among the novice and low-skill workers. Similarly, Noy and
Zhang (2023) find that having access to ChatGPT helps improve the productivity of writing
professionals, by increasing the quality of outputs as well as by reducing the amount of time
required to produce them, with the benefits being the largest for low-ability workers.
While the results of this literature suggest a promising role for GenAI to boost productivity, in the
context of LAC and emerging economies more broadly, there are important reasons to be cautious.
First, there are good chances that such initial macroeconomic projections are too optimistic and
based on oversimplified models. As shown by Acemoglu (2024), when the tasks classified by
Eloundou et al. (2023) as exposed to GenAI in the context of US-based occupations are linked to
their actual impact on GDP, the average task-level savings and the economic viability AI
deployment (Svanberg et al., 2024), the estimated impact amounts to a modest 0.71 of additional
Total Factor Productivity at the end of a 10 year period. Accounting for hard-to-learn tasks drops
this estimate to 0.55 percent of TFP, corresponding to an additional GDP growth due to AI at 0.92
percent over 10 years. In addition, the actual impact on productivity in specific occupations might
be largely dependent on how these technologies will be implemented at the workplace. For
example, Doellgast et al., (2023) suggest that productivity benefits might be less consequential if
the new AI tools are applied mainly for worker control, thereby limiting creativity and the
opportunities for larger value added through innovation in products and services. Acemoglu (2024)
also demonstrates that the final impact on productivity largely depends on the type of new tasks
that will emerge due to adoption, and that some of such new tasks might either not contribute much
new economic value or produce outright “public bads” that can be wrongly accounted as part of
GDP growth based exclusively on their monetary value. 4
Second, the rate of adoption and exposure to GenAI is likely to be slower in developing countries
where fewer workers are using digital technologies than in their richer counterparts. More
specifically, two individuals with the same occupation could have very different levels of GenAI
exposure if one of them uses a computer or internet at work, while the other one does not. As
4
E.g., tasks related to dealing with the increasing complexity and costs of network security or content manipulation.
6
shown in Figure 3, internet access in LAC countries varies from anywhere below 50 percent to
over 90 percent of the population, with the digital divide clearly correlated with income-based
differentials among countries. One of the main objectives of our paper is to quantify the limiting
effects of such digital gaps in the LAC region, which also provides a proxy for the challenges that
regions with even lower levels of income and digital infrastructure are likely to face. The
underlying assumption of our approach is that having access to a computer and internet at work is
a minimum requirement for drawing productivity benefits from GenAI tools. Consequently,
workers without such digital basics will simply be excluded from any form of productivity gains
that GenAI could offer in the professional context.
Figure 3. Internet coverage vs per capita income: global and LAC
Third, beyond the hard infrastructure, software costs are likely to impact the economic viability of
adoption in developing countries. Basic licensing of such products as ChatGPT or MS Co-pilot
can range around 20-30 USD per user per month, which can be significant, especially if applied to
a range of workers in one company. Costs of enterprise-level solutions, either based on simple API
integration or more complex proprietary AI-systems can be significantly higher. In countries with
high informality, including those of the LAC region, such costs are prohibitive for many small
enterprises, which exist outside the reach of any public support schemes of rapid technological
adoption. In the high-income context of the US, Svanberg et al. (2024) calculated that among all
occupations with a high theoretical AI exposure, the current costs of automation with computer
vision technology would make most businesses abstain from immediate implementation. Even a
rapid decline in the costs of such technologies would still make the actual deployment a gradual
process. 5 Given the lower income levels and higher digital limitations, in the LAC countries such
a process should be proportionally slower, especially in smaller and enterprises and in the informal
sector.
Fourth, workers need a minimum level of foundational skills to fully reap the benefits of this
technology (Autor, 2024), and such skills are likely to be scarcer in developing countries
(OECD/PIAAC, 2019). In the LAC region, the gap in the stock and quality of human capital
5
Brynjolfsson et al. (2021) show that technology adoption typically follows the J-curve, of which the flat part can
extend over multiple years (Acemoglu, 2024).
7
compared to developed nations was already significant before 2020 (Bakker et al., 2020) and the
school closures during the COVID-19 pandemic widened it even more (Schady et al., 2023). The
recent experimental studies with GenAI are focused on very specialized groups of workers
engaged in complex tasks, who are likely to be at the top of digital skills distribution across most
developing countries. That is, the existing evidence of GenAI reducing skills inequality within
these groups cannot be directly extrapolated to the whole economy or assumed to apply across the
broader spectrum of the labor force typical in developing countries. In the case of LAC, the large
informal sector is also likely to contribute to weak skills transferability across many occupations,
with fewer opportunities for on-the-job training or state-supported skills development schemes
that can be found in HICs.
Fifth, the results of these recent experiments and macroeconomic models do not consider general
equilibrium or second order effects on employment. For example, while increased productivity
may bring employment and wage gains in sectors facing a consumer demand that is growing
rapidly, that may not be the case for sectors facing a more stable consumer demand (Autor, 2024).
The nature of these second order effects is likely to be different across countries. In developing
economies with a large fraction of the workforce in the informal sector, and where technology
adoption and private sector investment are typically concentrated among a small share of formal
firms (Cirera and Cruz, 2022), workers displaced from formal sector jobs may face more
challenges finding high quality jobs than their counterparts in high-income countries. While
detailed macroeconomic modelling of such effects is beyond the scope of our study, the estimates
of jobs’ exposure to GenAI presented in this paper provide a profile of the socio-economic groups
more likely to experience the first-order impacts.
Historically, together with Sub-Saharan Africa, LAC is one of the most unequal regions in the
world (World Bank, 2016a), with levels of income inequality strongly influenced by the changes
in the structure of the labor market (Azevedo et al., 2013). Concerns about the impacts of new
technologies on inequality in LAC are consistent with broader empirical evidence about the effects
of recent waves of technological change on labor demand, which tended to be skill-biased and to
widen the gap between low- and high-income workers (Acemoglu and Restrepo, 2022; Autor et
al., 2008). Acemoglu's (2024) most recent modelling of GenAI outcomes on wages and inequality
also suggests that in nearly all theoretical scenarios, the deployment of this technology at the
workplace is likely to increase the inequality between capital and labor, and result in higher income
inequality between different demographic groups, with particularly negative consequences for the
incomes of low-education women in the US. 6 In the case of LAC countries, Dutz et al. (2018)
conduct a comprehensive discussion of the several challenges for the region in terms of digital
technologies and how inclusive they might be, looking at diverse case studies of technology
adoption in Latin America. In this regard, the distributive impact of the AI adoption depends
strongly on how the effects of increased productivity and output could overcome labor
displacement conducted by substitution of technology for workers. Although the use of AI
technologies in Latin America remains still very low, recent empirical evidence shows LAC labor
patterns more consistent with the skill-biased technological change hypothesis than the job
6
Albeit with smaller wage effects than the previous waves of automation (see Acemoglu and Restrepo, 2022).
8
polarization model 7 (Brambilla et al., 2023; Messina et al., 2016; Messina and Silva, 2018). Similar
conclusions arise from the literature on employment and automation in the rest of the developing
world (Das and Hilgenstock, 2022), although Maloney and Molina (2016) find some evidence of
incipient polarization in Brazil and Mexico.
To further theorize the potential effects of GenAI diffusion on inequality in the region, Figure 4
presents the most recent breakdown of LAC occupations by the highest, 1-digit level of ISCO-08,8
revealing visible differences in the employment structures across genders.
Figure 4. Occupations in the LAC region, by ISCO 1-digit and gender
Note: The breakdowns are presented as a share of male and female employment separately and calculated as a mean share
of employment across the countries in each income bracket, based on ILO modelled estimates (ILO, 2023a).
For men, the largest share of employment is in the elementary, agricultural, forestry and fishery
work, followed by craft and related trade workers. For women, the largest employment categories
concern service and sales work, followed by elementary jobs. Among the “Service and sales
workers”, the pattern is very similar across country groups, with male employment dominant only
in protective services, and female employment having much higher shares in personal care, sales
and personal service work. A more detailed analysis at ISCO-08 2-digit level 9 reveals that –
excluding IT, science and engineering professions – women are significantly more represented
across all professional categories, with particular prominence in teaching, health, business
administration and legal, social and cultural occupations. This trend extends into clerical work and
amplifies in line with countries’ income status. This warrants attention, since recent research has
7
The skill-biased technological change (SBTC) hypothesis suggests that technology benefits skilled workers,
increasing demand for high-skill jobs and widening wage inequality. In contrast, the job polarization model posits
that technology creates more high-skill and low-skill jobs, reducing middle-skill job opportunities and hollowing out
the middle class.
8
Elementary occupations are grouped together with agricultural, fishery and forestry work (96).
9
Plot A1 in the appendix shows a full breakdown of occupations at a 2-digit level of ISCO-08, by country and gender.
9
identified clerical and professional job categories as being more exposed to the risks of automation
with GenAI (Cazzaniga et al., 2024; Gmyrek et al., 2023; Ozdeneron, Hakki, 2023; WEF, 2023),
with pre-GenAI regional assessments also classifying female-held jobs in LAC as being at a higher
risk of automation from digital technologies (Egana-delSol et al., 2022).
In accordance with the technical documentation of ISCO-08, such differences in the occupational
structures also correspond to varying levels of skills and educational attainment, as clerical support
workers, technicians and professionals are typically classified in the mid- to high-skill level
brackets (ILO, 2023b). Given that educational attainment and earnings gaps across skills groups
have been important drivers of income inequality in LAC (Azevedo et al. 2013), the impact of
GenAI that follows the existing labor market structures would likely also have an effect on the
overall income inequality. In the best-case scenario, GenAI would boost the productivity of lower-
skilled workers in the exposed occupations, allowing them to access higher incomes and therefore
leading to a more broad-based income distribution. In the worst-case scenario, the technological
transition could result in the automation of largely female-held jobs in the clerical, technical and
professional occupations, while the opportunities for new GenAI-augmented jobs could be limited,
given the high concentration of current employment in elementary occupations and in the informal
sector, where technology adoption and private sector investment are low. To better understand
how the first-order effects of GenAI may affect inequality, this study provides a detailed profile of
the socio-economic groups most exposed to this technology.
Finally, we acknowledge that the final outcomes of the technological transition process will also
be largely dependent on the existing and future policy frameworks in the region. While the analysis
of country-level polices and legal frameworks is beyond the scope of this regional study, the
detailed country-level statistics that we make publicly available alongside this publication can
serve as useful inputs to the discussions underpinning such policy responses. 10
10
Access to detailed data at: https://pgmyrek.shinyapps.io/AI_Data_Portal_Research/.
10
3. Methods
3.1. Occupational exposure to GenAI
We combine multiple datasets to estimate occupational exposure to AI, leveraging the distinct
advantages inherent in each dataset to ensure a comprehensive analysis.
We use the AI exposure scores at the 4-digit ISCO-08 level from AI scores from GBB (2023) as
the principal indicator of occupational exposure to GenAI. GBB scores were developed based on
the technical documentation of ISCO-08, which contains a list of typical tasks for each of the 436
detailed occupational groups at the most detailed, 4-digit level, and which forms the basis on which
national statistical labor survey reports are linked to the internationally comparably ISCO-08
standard at the ILO. GBB build on the findings of Eloundou et al. (2023), who demonstrate a close
alignment of GPT-4 predictions with a survey of 70 AI experts on the potential of automating
occupational tasks with LLMs, and more broadly on Bubeck et al. (2023), who provide extensive
tests of the model’s capabilities and demonstrate its capacity for elaborating logical links between
items, resolving complex tasks and providing justifications for its decisions. Using the Application
Programming Interface (API) of GPT-4, the authors designed a sequential call that loops over each
of 3,123 tasks in that documentation and requested the model to assess the technical feasibility of
performing a given tasks with GPT-4 or LLM technology of similar capabilities. The model is
asked to rate tasks on a scale of 0 to 1, with 1 representing the possibility of performing a given
task by the LLM in full autonomy form a human operator, and to elaborate a written justification
of each score (no task received a score of 1). The scores and justifications are then reviewed for
consistency and stability of predictions over time, with the written justifications reviewed by
humans. Tasks with scores above 0.8 (high possibility of automation) are transformed into
embeddings, with a semantic clustering algorithm applied to identify the major groups of such
tasks, which are subsequently reviewed by humans.
Task-level scores for each occupation are used to calculate the mean score and the standard
deviation (SD) for each occupation. These two moments of distribution are subsequently used to
elaborate a theoretical framework for further classification of scores. Occupations (i) with a high
mean (µi > 0.6) and a high difference between the mean and SD (µi - σi > 0.5) are classified as
jobs with a high automation potential. Occupations with a low mean score (µi < 0.4) and a high
sum of the mean and SD (µi + σi > 0.6) are considered to have a high potential for augmentation,
meaning that while some of their tasks could be automated, the human role remains crucial for the
majority of their tasks. Occupations between these two categories are classified as “the big
unknown”, since, depending on the progress of technology and the use of adjacent technological
applications (e.g. LLM-based agents), they could fall closer to automation or augmentation.
Remaining occupations are classified as not affected, with the understanding that GenAI in its
current form would have minimal or no impact on their tasks. The scores and individual task
distributions are visualized by the authors through a publicly available interactive app:
https://pgmyrek.shinyapps.io/AI_Data_Portal_Research/. While GBB calculate separate scores
for high- and low-income countries, the results are very similar and thereby only the high-income
ones are used for all countries regardless of their income levels. See Appendix for a comparison
of GBB scores to Felten et al. (2023) and Prytkova et al. (2024). See Gmyrek et al. (2023) for a
detailed description of the score generating process.
11
We also consider alternative scores that could be used for this purpose, in particular the ability-
based scores developed by Felten et al. (2021, 2023a, 2023b, 2018), based on the US O’NET
classifications, recently linked to ISCO-08 4-digit occupations by (Cazzaniga et al., 2024), as well
as the patent-based scores of exposure to digital technologies by Prytkova et al. (2024), from which
a set of technological groups relevant to GenAI could be isolated. Having compared these
alternatives, we find that Felten et al. scores are quite aligned with GBB in broad terms, except for
catching a much wider group of Managers and Professionals as highly exposed to AI
technologies. 11 Since such broad coverage seems somewhat unrealistic in the context of many
developing countries, we opt for the scores of GBB, which provide a direct link to ISCO-08
documentation and focus exclusively on GenAI.
This choice is further reinforced by the arguments recently advanced by (Nurski and Ruer, 2024),
who find that task- (GBB) and ability-based (Felten, 2023) scores render similar general results in
the European context. The authors suggest the task-based analysis using GBB scores is particularly
advantageous for evaluating employment impacts, since task bundles are better at representing the
daily reality of occupations (see Autor, 2015), and considering the share of affected tasks offers
more scope to separate the potential for job transformation or displacement due to technological
advancements. Indeed, task-based approaches have been widely used in the literature for this type
of analysis (Restrepo, 2023, for concrete examples see Acemoglu and Restrepo, 2022, 2018; Frey
and Osborne, 2017;), including the recently increasing use of AI-generated task-level scores
(Acemoglu, 2024; Eloundou et al., 2023), used as a blueprint for the GBB scoring method.
In step 1, we tag occupations at 4-digit level in ISCO-08 into three categories established by GBB:
“automation potential”, “augmentation potential” and “the big unknown”. We then rely on the ILO
harmonized microdata collection 12 to obtain the shares of employment at 4-digit level occupations
for 18 countries for each of these three AI exposure categories (Figure 7). We also calculate the
shares that such exposed occupations make up in the higher, 2-digit level of occupational
classification. From this step, we switch to the harmonized household surveys from the Socio-
Economic Database for Latin America and the Caribbean (SEDLAC) to calculate AI exposure
across and within 16 Latin American countries. 13 Our sample of SEDLAC data consists of about
900,000 individual survey observations, with details by country provided in the Appendix (Table
A1).
The advantage of SEDLAC database is that it contains a host of harmonized variables at the
individual level, including the income aggregates used to measure poverty, as well as demographic
11
See Figures A2 and A3 in the Appendix for a quick visual comparison of GBB scores to Felten et al. (2023) and
Prytkova et al. (2024). Felten et al.’s scores cover a wider range of AI than GenAI covered by GBB. Prytkova et al.
(2024) focus on tech abilities in patents, which can be far from readiness for market-level deployment. A detailed
analysis of these scores at the 4-digit occupational level is available upon request.
12
Calculations of 2-digit employment from ILO Micro data repository by David Bescond, ILO STATISTICS.
13
SEDLAC is produced by the University of La Plata’s center for Distributional, Labor and Social Studies (CEDLAS)
and The World Bank’s Equitable Growth, Finance and Institutions LCR-POV-Poverty and Equity Group (ELCPV).
This project aims to improve the comparability of social and economic statistics across 25 countries in the Latin
American and the Caribbean (LAC) region. This involves the harmonization of household survey variables in eight
categories: income, demographics, education, employment, infrastructure, durable goods, services, and aggregate
welfare.
12
characteristics and labor market outcomes. In step 2, we impute the AI exposure scores from GBB
to individual respondent data in SEDLAC, using the ISCO-08 occupation reported in the
household survey. Such imputation is straightforward for the 8 countries with 4-digit ISCO-08
occupations in SEDLAC, and for which we can directly compare the calculations to the estimates
from the ILO as an additional validation measure (Figure 5). In contrast, there are 8 countries in
SEDLAC with 2-digit ISCO-08 scores where the imputation is less obvious and depends on other
circumstances. We have two types of such cases.
Figure 5. Coverage of ISCO-08 4-digit microdata in SEDLAC (WB) and ILO harmonized microdata collection
When we have the 4-digit ISCO-08 employment structure from ILO for the same country, we use
the estimates of the shares of exposure at the 2-digit level calculated in step 2 above. 14 When we
do not have the 4-digit employment structure from a different source for the same country, we use
that of a “similar” country. 15 These country similarities are defined in step 3, by applying a
hierarchical clustering algorithm to several country-level characteristics including the full
breakdown of 2-digit ISCO-08 employment shares, GDP per capita (PPP) and total population
(Figure 6).
14
This concerns Brazil, Colombia, Costa Rica and Mexico.
15
This concerns Argentina, Bolivia, Guatemala and Nicaragua.
13
Figure 6. Hierarchical clustering based on ISCO 2-digit shares, GDP(PPP) and total population
In step 4, we impute the estimated shares of automation, augmentation and the big unknown to
individual responses at the 2-digit ISCO-08 occupation level in SEDLAC. Having a data frame
with 2-digit level shares enables aggregation of individual responses by main categories of interest
captured in SEDLAC microdata. We focus on gender (male, female), area (rural, urban), age (15-
14, 25-34, 35-44, 45-54, 55-64), education (low, medium, high), poverty status (non-poor, poor),
income quintiles (Q1 through Q5), formality (legal, productive), labor relationships (employer,
salaried employee, self-employed, family worker without salary) and sector of economic activity.
The shares of exposure are calculated in such a way that automation, augmentation, big unknown
and other occupations add to 100 percent within each category. This means that we can interpret
such results as a share of employment in each type of AI exposure within each grouping category
(for example, shares of automation, augmentation, big unknown and other occupations among
people with low education or among those belonging to the age bracket of 35-44). Table 1
describes these variables in more detail.
Table 1. Distribution of AI Exposure by Demographic and Socioeconomic Categories in SEDLAC Data 16
Variable name Description
Education Low: fewer than 9 years of education
Middle: 9 to 13 years of education
High: 14 or more years of education
Poverty An individual is considered poor (non-poor) if they live in a household whose
income per capita is below (above) the poverty line for upper middle-income
countries (US$ 6.85-a-day in purchasing power parity terms)
Income quintiles Q1 through Q5, by whether the individual’s household income per capita is in said
quintile.
Formality (legal) A salaried worker is informal if they do not have the right to a pension linked to
employment when retired
Formality (productive) An individual is considered an informal worker if they belong to any of the
following categories: (i) unskilled self-employed, (ii) salaried worker in a small
private firm, (iii) zero-income worker. Unskilled workers are all individuals without
a tertiary or superior education degree. Small firms are those with 5 or fewer
employees. These criteria and definitions refer to individuals’ main job.
16
For more details see https://www.cedlas.econo.unlp.edu.ar/wp/wp-
content/uploads/Methodological_Guide_v201404.pdf.
14
Sector of economic Primary sector
activity Low-tech manufacturing (food, beverages, tobacco, textiles and clothing)
Other manufacturing
Construction
Retail, restaurants, hotels and repairs
Utilities, transport and communications
Banking, finance, insurance, professional services
Public administration
Education, health and personals services
Domestic service
3.2. Use of a computer at work
The method applied so far enables detailed insights into country-level data on exposure of
occupations to GenAI, with further breakdowns by demographic and socioeconomic
characteristics of the affected groups (Table 1). At the same time, the variation in AI exposure
across and within countries is only driven by the variation in the occupational structures, because
the same occupation in different countries uses the same score of GenAI exposure. To address this
limitation, in the next step we introduce cross-country variation of occupation-level scores, by
accounting for the variability in the use of computer equipment in the same occupation located in
different national contexts.
We first proceed by imputing the GenAI exposure measures at the 4-digit ISCO08 to the microdata
from the Programme for the International Assessment of Adult Competencies (PIAAC) collected
by the OECD. These surveys include rich information on detailed tasks carried out by people at
work, such as whether workers use a computer (and internet) 17 at work. Using this binary indicator,
we split each group of GenAI exposure into those who use a computer at work, and those who do
not. Not using a computer at work means that even if the worker is in an occupation that is exposed
to GenAI augmentation, such potential productivity gains are unlikely to realize given the lack of
access to digital infrastructure. We first implement this exercise for the four countries in the LAC
region (Chile, Ecuador, Mexico and Peru) and two developed economies (Slovenia and New
Zealand) included in the PIAAC dataset. 18
17
For the main results presented in the paper, we use variable “g_q04”, which contains the response to the following
question: “Do you use a computer in your job?/Did you use a computer in your last job?”. We also do robustness
checks by creating a binary variable equal to 1 when the worker uses both a computer and internet at work, by using
the variables “g_q05a”, “g_q05c”, “g_q05d” and “g_q05h”, which contain information about the frequency (i.e.
“never”, “less than once a month”, “less than once a week but at least once a month”, “at least once a week but not
every day”, or “every day”) of internet use for mail, work related information, conduct transactions and participate
in video calls, respectively. More specifically, when the worker responds that he or she “never” uses the internet for
any of those four reasons at work, we consider that the worker does not use internet at work. If the worker responds
that he or she uses the internet at work for any of those purposes with any frequency other than “never”, then we
consider that the worker uses internet at work. Simultaneous use of computer and internet presents a more restrictive
condition, which results in greater digital gaps. These calculations at the country level are shown in Figure A5. More
detailed statistics are available upon request.
18
While PIAAC includes several developed economies, we did not choose them for this initial step of the analysis
because their surveys were either collected several years earlier (2011-2012) and the adoption of digital technologies
15
Since there are only four Latin American countries in PIAAC, we extrapolate the measures of
computer use at work from PIAAC to the full set of countries in the SEDLAC database. In
particular, we estimate a predictive model for the probability of computer use at the individual
level using the full set of countries in PIAAC 19 and independent variables that are available both
in the PIAAC and SEDLAC databases. 20 We then use the estimated model and the set of
independent variables to predict the probability of computer use in the SEDLAC database. More
specifically, we first estimate the following Logit model in PIAAC:
������������ ������������
Pr (������������������������������������������������������������������������������������������������������������,������������ = 1) = ������������ (������������������������������������������������������������ ,������������ , ������������������������������������������������ ,������������ , ������������������������������������������������������������������������������������,������������ , ������������������������������������������������,������������ , GDP������������ , ������������������������������������������������������������������������������������������������������������ , ������������������������������������������������������������������������������������������������������������������������ ) (1)
Where ������������������������������������������������������������������������������������������������������������,������������ is a binary variable equal to 1 if individual ������������ in country ������������ uses a computer at
������������ 21 ������������
work; ������������������������������������������������������������ ,������������ is a vector of 39 dummy variables for each 2-digit ISCO08 occupation ; ������������������������������������������������ ,������������ is
a vector of 4 dummy variables indicating age groups; ������������������������������������������������,������������ is a dummy variable equal to one for
High School graduates; GDP������������ is the log of GDP per capita in 2017 US$ PPP; ������������������������������������������������������������������������������������������������������������ is the rate
of internet users per 100 people, and; ������������������������������������������������������������������������������������������������������������������������ is the number of fixed broadband subscriptions
per 100 people. Since the reference year of the PIAAC surveys varies by country, we use the
corresponding year of the country-level variables (i.e. GDP, internet and broadband). These
country-level variables are helpful to capture the link between the economy-wide level of digital
and economic development with the level of computer use at work.
In the next step, we use the estimated equation (1) to predict the probability of using a computer at
work at the individual level in the SEDLAC database. 22 The probability of not using a computer at
work is simply 1-Pr(computer=1). When choosing the reference years of the country-level
variables of the model, we use the reference year of the SEDLAC surveys. Then, the probability
of an individual being exposed to AI and of using a computer at work will be the multiplication of
both individual probabilities. For example, take the case of a group of workers (e.g. workers with
high education) that, on average, has a 0.23 GenAI augmentation exposure probability (i.e. the
average of the binary exposure measure at the 4-digit level). Then let’s assume that, based on the
individual predictions from the Logit model, on average, individuals in such group have a 0.7 (0.3)
likelihood of using (not using) a computer at work. As a result, we conclude that workers in such
group have a 0.161 (=0.7 x 0.23) probability of being exposed to GenAI augmentation and of using
a computer at work, while they have a 0.069 (0.3 x 0.23) probability of being exposed to GenAI
augmentation and of not using a computer at work. 23
increased dramatically since then. In addition, some developed countries with more recent surveys do not have
ISCO 08 information at the 4-digit level (e.g. United States). In contrast, New Zealand and Slovenia’s surveys were
collected during the second round (2014-2015), which is the same timeframe of Chile’s survey, and closer to the
timeframe of Ecuador, Mexico and Peru (2017).
19
There are 38 countries with publicly available PIAAC microdata, see https://www.oecd.org/skills/piaac/data/.
20
A similar exercise was implemented by Garrote et al., (2021) to adjust working-from-home measures by internet
access rates.
21
While there are 43 ISCO08 2-digit level categories, we drop from the sample the categories 01 (Commissioned
Armed Forces Officers), 02 (Non-Commissioned Armed Forces Officers) y 03 (Armed Forces Occupations, Other
Ranks) since they are not available in PIAAC.
22
Table A2 in the appendix contains the estimated coefficients.
23
The logic is the same when the measures of GenAI exposure are imputed at the 2-digit ISCO08 level, which are not
binary but continuous measures between 0 and 1. That is, the exposure to GenAI augmentation within a group of
workers would be the average of the continuous exposure measure.
16
In the final step, we calculate measures of AI exposure across the same socio-demographic
characteristics as summarized in Table 1, this time with a simultaneous breakdown by computer
use at the workplace for each of these characteristics.
17
4. Findings
4.1. Cross-country comparisons of the levels of exposure
Since we apply the same conceptual framework of exposure as proposed by GBB (automation,
augmentation and “the big unknown”) we can calculate the total share of employment exposed to
GenAI by adding those three categories in each country. Figure 7 presents the ranking of individual
countries, with total exposure ranging between 26-27 percent in Barbados, Ecuador and Bolivia,
up to 37-38 percent in Uruguay and Costa Rica respectively. 24
We strongly emphasize that this way of presenting results should not be equated with a statement
that “up to 40 percent of employment in LAC is exposed to automation”. Quite to the contrary,
among these fairly large shares of potential exposure, only a small portion of occupations –
between 2 to 5 percent, depending on country context – falls into the potential of full automation.
Augmentation figures are consistently higher, ranging between 8 and 14 percent across countries.
In addition, the category of the big unknown is rather large, suggesting that in the medium-term,
these estimated shares could change quite significantly, as occupations from the big unknown
might move into the automation and augmentation categories, depending on whether the
technology is applied in a more task-automating manner, or whether it generates new
complementary tasks (Acemoglu, 2024). In the short term, however, the proportion of employment
in the LAC countries that have the potential to completely disappear to GenAI automation is
significantly lower than the jobs that could be transformed by this technology. It is also important
to emphasize that these cross-country comparisons regarding GenAI exposure rely on the
assumption that the same occupation has a similar exposure in, for example, a high-income country
and Guatemala. Some empirical work suggests that this assumption may not always hold, since
the tasks of each occupation tend to change with economic development (Caunedo et al., 2023;
Lewandowski et al., 2019; Lo Bello et al., 2019). 25 This paper tries to mitigate this issue in Section
4.2 by adjusting these exposure measures with measures of computer use at work.
24
Argentina‘s labor force survey covers only larger cities, which is why the exposure to GenAI is significantly larger
than for other countries (44%). It was removed from the plot to avoid confusion. Data for Grenada, Guyana,
Montserrat, and Suriname are based on ILO micro data, all other countries based on SEDLAC. Total employment for
Grenada and Montserrat based on 2020 surveys, for all other countries based on ILO modelled estimates for 2023.
25
However, since the patterns of task specialization across countries differ along several dimensions, it is difficult to
assess whether our method tends to under- or over-estimate the exposure to GenAI in the developing world. For
example, Caunedo et al (2023) find that occupations tend, on average, to be more intensive in non-routine interpersonal
tasks in richer countries. If such tasks are less prone to be automated by GenAI, then our method would be under-
estimating the exposure to GenAI in developing countries relative to developed ones. On the other hand, occupations
tend to be more intensive in non-routine manual tasks in poorer countries. If such tasks are less likely to be automated
by GenAI, this would imply that our method would over-estimate the exposure to GenAI in developing countries
relative to developed ones.
18
Figure 7. Total exposure to GenAI by country 26
For reference, we conduct the same estimation for all HICs, calculating the mean level of exposure
in the three categories (Figure 7). 27 In line with the theoretical arguments considered in section 2,
the variation in LAC countries’ overall exposure is strongly related with the income levels, since
labor market structures in higher-income countries tend to contain proportionally more
occupations that are likely to come into direct interaction with GenAI technology. Accordingly,
wealthier countries of the LAC region come closest to the HIC reference category, where total
exposure reaches 43 percent. Beyond this general income effect, important variability can be
observed in each category of exposure, as shown in greater detail in the sections that follow. 28 To
explore these differences, we examine the cross-country variation in the shares of occupations
within the two opposed categories of “automation potential” and “augmentation potential”, with
further breakdowns by demographic and labor market characteristics.
26
Total number of jobs based on the ILO modeled employment estimates (ILO, 2023a).
27
Group estimation based on the same method as GBB, data provided by David Bescond, ILO STATISTICS.
28
See Figure A4 in the Appendix for country-rankings based on each type of exposure.
19
Figure 8. Automation potential - detailed breakdown of socio-economic characteristics
Note: Black dots in the chart represent the country-level exposure measures for each corresponding socio-economic group. The red dots indicate
the average level of exposure.
Figure 8 presents these breakdowns for the category of automation, with the means of country-
level shares marked in red. On average, the share of female-held jobs exposed to automation is
double of the share of jobs held by men. The dispersion of country-level points is higher among
female-held jobs, with some important outliers representing an even larger difference between the
(lower) exposure of men to automation and the (higher) exposure of women. Figure 8 also reveals
a difference in terms of job location, with a significantly higher share of jobs potentially exposed
to automation located in the urban areas. Furthermore, the shares of such occupations are highest
among young workers (15-24 and 25-34) and those with medium and, especially, high education.
The average shares of employment in this category are high among the non-poor and increase with
household income levels in a somewhat linear pattern, and that the exposure to automation is
highest among workers with formal jobs and with salaried employment contracts. Finally,
disaggregation by sector shows that the highest shares of jobs with exposure to automation are
found in banking, finance and insurance services, followed by public administration and defense.
In other words, if one was to describe a person exposed to the potential of automation in the LAC
region, the defining features of that average profile would be: “female, urban, young, medium-to-
20
high level educated, non-poor, with a relatively high income and a formal employee job in banking,
finance and insurance businesses, or in the public sector”.
Figure 9. Augmentation potential - detailed breakdown of socio-economic characteristics
Note: the chart shows the country-level exposure measures for the corresponding socio-economic group. The red dots indicate the average level of
exposure.
Figure 9 mirrors the analysis, with a focus on the exposure to augmentation. While several trends
are similar to the automation exposure, important differences also emerge. First, the exposure to
augmentation shows significantly less gendered effects: while the share of exposed female
employment is higher than male employment across all countries, these differences are
significantly smaller and, in some cases, negligible, with a higher cross-country variation. The
same applies to the variables of age, with a slightly higher mean share of jobs in the 25-34 and 35-
44 bracket, but no dominant pattern is observed across countries. 29 In other words, when it comes
29
Such differences are more visible within country-specific data discussed in the following section 4.3.
21
to gender and age, the benefits of potential transformation are more equally distributed across these
dimensions than the exposure to the risk of automation. The trends across several other
characteristics are similar to the automation exposure, with urban, higher-educated, non-poor,
formal occupations and higher income brackets corresponding to relatively higher augmentation
potential. An important difference, however, concerns the employment relationships and sectors.
Whereas automation was most prevalent among employees, the potential of augmentation is
relatively higher among employees and the self-employed, with significantly higher shares of
employment observed in education, health and personal services. The exposed self-employed
occupations with the largest share of jobs cover a wide range of the skill spectrum: low (e.g.
elementary workers not classified elsewhere), middle (e.g. hairdressers, market salespersons,
motorcycle drivers) and high (e.g. architects, real estate agents, photographers, musicians).
In summary, having compared the trends across countries and socioeconomic groups, we observe
some common patterns regarding overall exposure to GenAI, with important variability across
types of exposure. People with higher education and incomes, in urban areas, with jobs in the
formal sector, and younger tend to have higher levels of exposure to GenAI, regardless of the type
of exposure. The strongest differences in the type of exposure appear across gender, employment
status and sectors, with women and salaried employees in banking, finance, insurance services and
public administration showing higher exposure to automation, while augmentation exposure is
most dominant among salaried and self-employed and in the sector of education, health and
personal services. 30
4.2. Impact of digital infrastructure on the potential of transformation
The previous section provided a detailed analysis based on the assumption of equal exposure
among all countries within a given ISCO-08 2-digit category. In this section, we apply the
calculations that predict the use of a computer at work and therefore introduce variation at the level
of individual occupations within countries. 31 Since access to a computer is a crucial starting point
for the use of GenAI technologies, by applying this additional criterion, we separate the theoretical
exposure of occupations to GenAI from the practical feasibility of such a transformation in a given
country context, as a function of available digital infrastructure. We focus this part of analysis on
three principal dimensions: country-specific digital characteristics, exposure type and gender. This
worker-level approach suggests that the workers’ skills are crucial to determine the digital divide
and the effective exposure to GenAI. But it is worth noting this gap could also be determined by
firm characteristics or AI investment decisions. Although there is no clear agreement on the effects
30
These general patterns also hold when conducting a conditional analysis. In particular, we estimate OLS
regressions to simultaneously account for the different drivers of exposure to GenAI by pooling the microdata for all
16 countries (Table A3 in appendix). To consider the sampling frame of the household surveys while at the same
time giving the same weight to every country, the population weights are normalized so that they add up to 1 within
each country. We do not use the population weights in their raw version because the results would be driven mostly
by large countries such as Mexico and Brazil. The main differences from the unconditional figures analyzed above
are that women are less exposed to augmentation than men, and that rural and urban workers are equally exposed
when controlling for other factors.
31
See Section 3.2 for technical details.
22
of AI technologies on employment (Babina et al., 2024; Restrepo, 2023), there is a strong
consensus that the technological divide could be widened among firms of different sizes. 32
We start by examining the estimates based on PIAAC surveys, which features four LAC countries
that we compare to Slovenia and New Zealand. This helps us assess the results based on the same
survey method and corresponding years of data collection across several countries before we
proceed with more complex probabilistic modelling between the PIAAC and SEDLAC datasets.
Figure 10. Jobs with augmentation potential and access to computer at work, based on PIAAC data
Figure 10 shows that the four LAC countries in the sample have a similar share of employment
representing jobs with augmentation potential as Slovenia and New Zealand, with Chile’s total
share being the largest in this sample. However, when it comes to the use of a computer among
workers with these jobs, all four LAC countries take the lower end of the scale. In fact, in the case
of Mexico and Peru, the proportions are nearly inverse of those in Slovenia, as jobs with the
potential to be transformed by GenAI and using a computer at the workplace represent 5.1-5.4
percent of employment, compared to 8.5-8.7 percent of employment in the jobs that could be
transformed but do not have a computer connection. In the case of New Zealand, there proportions
are more than inverse, with 11.8 percent of employment in jobs with a computer and augmentation
potential and only 2.3 percent of employment in such jobs without a computer. While the overall
degree of digitalization has increased across countries over the past years, 33 PIAAC data seems to
provide one of the best available tools to capture the relative gaps between the highly digitalized
and less technologically advanced contexts. 34
32
In particular, the adoption of these advanced technologies concentrates in large and young firms, and is associated
with an increase in the demand for skills (Acemoglu et al., 2024, 2023; Autor et al., 2020; Babina et al., 2024;
Gutiérrez and Philippon, 2017; Webb, 2020).
33
For example, using the share of population as a proxy for digitalization, Chile moved from 76% in 2015 to 90% in
2022, Ecuador from 40% to 69 %, Mexico from 57% to 78%, Peru from 40% to 74%. During the same period, New
Zealand moved from 85% in 2015 to nearly universal use in 2022 (96%) and Slovenia from 73% to 88%. With internet
coverage frequently representing increase in the use of mobile phone access in developing countries, these increases
do not necessarily need to translate into the use of a computer in the workplace context.
34
The estimates in this study are based on the most recent data available prior to the 2022-23 second PIAAC survey
cycle. We contacted the OECD to request access to these surveys but were informed that the data could not be released
due to bilateral agreements with their member states. We plan to update our estimates with the new PIAAC data as
soon as they become publicly available.
23
As the next step, we proceed with the probabilistic model, imputing probabilities predicted based
on PIAAC data to our SEDLAC dataset. In the main analysis, we focus on the use of computer at
work, with an expanded analysis on the simultaneous use of a computer and internet presented in
the Appendix. In that sense, we focus on the lower threshold of the digital gap, as simultaneous
use of a computer and internet represents a more stringent criterion, which results in larger digital
gaps. Figure 11 presents a detailed breakdown of our aggregate calculations, with the share of
occupations in each country and exposure categories marked in orange in the case of jobs with an
available computer, and light blue for the jobs where a computer is not available at the workplace.
Figure 11. Exposure by country, exposure type and access to digital infrastructure
We can observe that, in the case of jobs exposed to potential automation, the shares of such jobs
that do not use a computer are generally very low across countries. In other words, most of such
occupations are already digitized. A notable exception applies to countries with lower income,
such as Nicaragua, Guatemala and Honduras, where around half of the jobs exposed to potential
automation are predicted to not use a computer. One way to think about this pattern is that, in
24
poorer countries, the lack of digital infrastructure might offer a temporary buffer from the risk of
imminent automation to some occupations in this category – a trend that likely extends to countries
with relatively lower incomes outside Latin America. The plot also re-confirms that such
automation-exposed jobs are disproportionately held by women.
The situation is visibly different in the case of occupations with augmentation potential. First, the
distribution of such jobs is more equal among women and men. In addition, the shares of jobs that
do not use a computer at work are also more evenly distributed across men and women. An
intuitive way of thinking about this part of Figure 11 is that the light blue zones (no computer)
represent the transformation potential that cannot be attained due to digital infrastructure
limitations. If one was to assume that such a transformation could translate into productivity gains
in those occupations, the zones without a computer can be seen as an unattainable potential
productivity gains. 35
Taking this analysis one step further, we can quantify these effects. For example, if we calculate
the unattained augmentation potential as an average value across the LAC countries (weighted by
total employment of each country), it corresponds to 6.24 percent of female employment and 6.22
percent of male employment. If we think of this gap as an average share of jobs within all jobs
with augmentation potential, it corresponds to a non-negligible 44 percent of such jobs held by
women and 50 percent of such jobs held by men. Applying these calculations to the 2023 ILO
modelled estimates of total employment in each country, we estimate that there are some 17 million
jobs among the 16 LAC countries in our SEDLAC sample that could, in theory, experience
additional productivity from the technological transformation with GenAI, but which will not be
in position to do so due to the lack of digital infrastructure. Some 7 million such jobs are held by
women and nearly 10 million are held by men (see Figure A6 in Appendix for country-level
breakdowns).
4.3. Within-country patterns
In this section, we present a sample of within-country analysis, using the example of Colombia
and Costa Rica (Figure 12). Detailed breakdowns by country are provided in the Appendix and
made available in a dynamic way on our online portal. 36
Within each country, the plot can be read as a detailed breakdown of a given category on the
vertical axis into the shares of employment represented in horizonal separations. For example, in
the case of Colombia, among women, 5.5 percent of employment is exposed to the risk of
automation, 11.3 percent of employment is in the augmentation category, and 22.5 percent in the
big unknown. The remainder, not presented in the graph, contains all other female-held
35
We also implement a robustness check where we use the variable of ”computer ownership at home”, as self-
reported by households in SEDLAC, to adjust for digitalization. In particular, we replace the imputed ”computer use
at work” variable for this measure, to check if the patterns prevail. As seen in figure A.5 in the appendix, while the
absolute values of exposure adjusted by digitalization are different across methods, the relative figures are very
similar. In particular, among jobs exposed to GenAI augmentation, the share of workers with a computer at home
tends to be higher in richer countries.
36
https://pgmyrek.shinyapps.io/AI_Data_Portal_Research/
25
occupations that do not fall into any of those three exposure categories. The dark shaded areas
represent jobs with a computer at work, while the light shading represent the jobs in each exposure
type that do not use a computer. We can observe that among the female-held jobs exposed to
potential automation, very few jobs do not use a computer, whereas among those with a
transformation potential, this digital constraint applies to about half of such jobs.
Figure 12: Exposure by country, type and detailed country-level characteristics
Figure 12 can also be read vertically, allowing for within-country analysis across characteristics
of both the worker and the job. Continuing with the example of Colombia, we can observe that the
share of jobs with high automation potential is higher among women (5.5 percent) than men (1.6
percent), and among urban (3.9 percent) than rural jobs (0.8 percent). The share of such
occupations is also highest among young people and decreases with age brackets, while it increases
with education levels and household income brackets. While these general trends have already
been discussed in the preceding section, the detailed breakdowns enable country-specific insights,
necessary for a broader reflection on adequate policy responses. In order to support such processes,
in addition to our paper, we make this detailed country-level data publicly available online.
26
4.4. Which occupations drive the effects?
To understand what drives these effects, in Figure 13 we focus only on occupations where at least
a quarter of all individual observations under a given ISCO 2-digit category falls into one of the
exposure groups. For example, taking “augmentation & computer”, we can observe that, in many
countries, over 50 percent of teaching professionals can be found in this category. The exceptions
concern lower income countries, such as Guatemala, Honduras and Nicaragua, where the majority
of teaching professionals are in the category of potential augmentation, but without access to a
computer at work. This suggests that the differences in digital development might impose
important limitations on the benefits of GenAI for the education systems in poorer countries.
Figure 13. ISCO 2-digit occupations by type of exposure and country (share of exposure > 25%)
27
Other occupations that could benefit from augmentation and already use a computer at work
concern legal, social and cultural professionals, and numerical and material recording clerks in
most countries, as well as health professionals, information and communication technicians and
some of the assemblers in countries with relatively higher incomes. In elaborating on these
categories, examples could include legal professionals like tax lawyers using GenAI for case
analysis, social workers employing case management software, and cultural professionals such as
digital archivists. Assemblers with a computer can include skilled workers like electricians who
may utilize GenAI-based diagnostic or instructional tools in higher-income countries.
Nevertheless, a larger share of the assemblers’ jobs falls into the category with augmentation
potential but no computer at work, alongside personal service and refuse workers. While such jobs
retain a central human component, in settings with a computer and internet access, some of their
tasks could benefit from the AI transformation. For example, personal service workers, such as
home health aides, could use a scheduling and client management software to enhance service
coordination, while companies hiring refuse workers could implement waste tracking systems and
route optimization software to improve efficiency and reduce environmental impact. 37 However,
due to the lack of digital infrastructure in LAC countries, such potential benefits would simply
remain out of reach to those job categories.
Figure 13 also demonstrates a variety of occupations falling into the “big unknown” category, with
a vast majority of such jobs already using a computer at work. This concerns business and
administration professionals and associate professionals, information and communication
technicians, customer service clerks, general and keyboard clerks as well as hospitality, retail and
other services managers. Decomposing these groups into more detailed occupations, it is easy to
illustrate why this category represents the zone between the potential of full job automation and
augmentation with generative AI. For example, business and administration professionals could
see GenAI software streamline complex data analysis, associate professionals might use it to
manage logistics more efficiently, while customer service clerks might rely on GenAI for support
with query resolution. As the technology evolves, the balance between augmentation and
automation of tasks might shift, potentially redefining some of the jobs in customer query roles
significantly faster and exposing them to a higher risk of full automation than other occupations in
this category. It should also be noted that a large share of customer query clerks is already found
in the category of high automation potential, alongside other groups of clerical occupations. The
fact that, in most cases, such jobs already use a computer at work further shortens the distance to
the potential full automation of these occupations, making potential shifts between the “big
unknown” and full automation more fluid.
4.5. Differential exposure across earnings levels
As the final step of the analysis, we look deeper in the relationship between labor earnings and the
degree of occupational exposure to GenAI, focusing exclusively on the occupations that report
already using a computer. To do that, we use the latest available micro data for each country and
37
This clearly assumes that such technologies would be adopted in a way that supports workers’ tasks, rather than
imposed to increase the level of algorithmic control and limit worker agency, which can have negative effects on
working conditions (Adams-Prassl et al., 2023; ILO, 2023c).
28
focus separately on income of employees and self-employed. To ensure comparability across
countries, we show the median income of each ISCO-08 2-digit level occupation as a percentage
of the median income of wage employees across all observations in each survey sample. 38
Figure 14. Earnings of occupations exposed to GenAI, by employment status (exposure above 25%)
Note: Incomes were calculated as the median income in local currency of each ISCO-08 occupation, based on
the latest available survey data for each country. They were then recalculated as a distance from each group’s
(employees/self-employed) median income and normalized as a share of the median income of employees in
each country, which provides a common reference point. Grey dots represent all occupations for which the
38
The median is favored in wage analysis as it resists skew from outliers. It represents the central tendency of a data
set and provides a clearer indication of typical income levels, especially in instances where income distribution is not
symmetrical.
29
sample size and existing data allow for calculation of the median income. Colored dots represent only these
ISCO-08 2-digit occupations where at least 25 percent of occupations in a country within a given 2-digit
category are estimated as exposed to automation, augmentation or the big unknown and have a computer at
work (theoretical readiness for GenAI-driven transformation).
In Figure 14, we first plot all country-level data points for LAC (grey dots), with the size of each
dot representing the employment shares. The dots below the horizontal reference line at zero
represent occupations with an income below the median income of wage employees in each
country. Conversely, dots above that line refer to jobs with income above the median income of
wage employees, with the lack dotted line representing the overall trend of higher incomes
accruing to jobs with a lower number in the ISCO-08 structure, that is, professional and managerial
positions.
Subsequently, we mark the jobs that are exposed to an immediate interaction with GenAI, that is,
occupations within the categories of automation, augmentation and the big unknown, which report
already using a computer at work. We highlight only these occupations in each country where at
least a quarter of jobs in a 2-digit category falls into one type of exposure, with exposure types
marked in colors and the share of employment reflected in the size of each marker.
We can observe that, in the case of wage employees, nearly all exposed occupations are either
around or above the median income, whereas for self-employed we see more occupations with
below-median incomes. The exposure to automation is generally grouped around occupations with
an income between the median and an equivalent of two median incomes, with only some of the
information and communication jobs exceeding that threshold. Occupations with a high
augmentation potential are generally grouped around the higher income bracket of somewhere
between 1.5 to 3 values of the median income of wage employees. The category of big unknown
is more spread across the income distribution, with some self-employed sales workers below the
median level and the top earners around three median incomes.
In the bigger picture, Figure 14 shows that most of the exposed categories concern jobs around
what could be defined as middle- and upper-middle income jobs, with hardly any occupations
showing significant exposure among the low-income occupations. In other words, the main thrust
of the first order effects of GenAI technologies can be expected among the people who already
have high incomes and who are in jobs requiring relatively higher skill levels, while the jobs of
the poor are quite likely to remain outside the immediate effects of this technological transition.
5. Final discussion
This study examines the exposure to GenAI within the labor markets of the LAC region, revealing
both widespread potential impacts and significant variability across different demographics and
sectors. Our findings indicate that a substantial proportion – between 30 and 40 percent of
employment in LAC – is exposed in some way to GenAI. This exposure is linked with the
economic status of countries, suggesting that income levels are a strong correlate of GenAI’s
impact on labor markets. However, it is crucial to note that such exposure does not imply
30
automation, and that for the vast majority of these jobs, the potential lies in transforming the tasks
that these occupations perform. Our estimates for the potential effects of automation in LAC
amount to 2 to 5 percent of employment depending on the country. These figures, while seemingly
modest, should not be trivialized as they represent individuals’ livelihoods that are at stake. In
addition, some of the jobs from the large category of “the big unknown” might move closer to
automation over time, as the technology and its applications to workplace tasks develop further.
Comparisons of our results to other studies are complicated, due to the significant differences in
the concepts applied, occasional lack of detailed data that would enable a more precise assessment,
diverging methods of presenting the findings, and the general scarcity of studies that cover non-
HIC countries (Comunale and Manera, 2024). For example, Eloundou et al. (2023) state that up to
80% of the US workforce could have at least 10% of their tasks replaced, while 19% of workers
could lose at least 50% of their tasks to LLMs – a finding that is hard to directly relate to our
framework, except for the similarity of a much stronger augmentation effect over automation.
McKinsey (2023) points to a similar group of “knowledge work” as being most exposed but focus
the analytical work on additional value generation through productivity increases, rather than on
direct effects on employment. WEF Future of Jobs (2023), even though global in scope, focuses
exclusively on large enterprises, pointing to clerical and administrative jobs among occupations
with the fastest expected declines. Goldman Sachs (2023), based on extrapolation of O*NET
occupations to emerging economies, suggests that “most jobs and industries are only partially
exposed to automation and are thus more likely to be complemented rather than substituted by
AI”. 39 According to our best knowledge, there are no prior studies with detailed insights to GenAI
exposure in the LAC region and our estimates of total potential exposure are generally lower than
the 40 percent estimated by Cazzaniga et al. (2024) for emerging economies. However, wealthier
LAC countries show exposure levels closer to this estimate.
Within this context, irrespective of country-specific differences, our estimations show that certain
characteristics consistently correlate with higher GenAI exposure. Specifically, urban-based jobs
that require higher education, are situated in the formal sector, and are held by individuals with
higher relative incomes are more likely to come into interaction with this technology. Moreover,
there is a pronounced tilt towards younger workers facing greater exposure, including the risk of
job automation, in particular in the finance, insurance, and public administration sectors. While
these groups might also be better positioned to reap the benefits of new technologies (Ananian et
al., 2006; Aubert et al., 2006 40; Cazzaniga et al., 2024), shedding well-paid, formal, skilled and
female-dominated jobs can hardly be a positive scenario for the already highly informal and
unequal economies of the LAC region. Our findings suggest that – at least as the first-order effect
– the middle class is the group whose jobs and earnings have the highest levels of overall exposure
to GenAI, with many possible directions that this transformation can take.
These findings suggest an important role for government interventions, aimed at minimizing
disruptions resulting from sudden job losses through job protection measures, and maximizing the
39
The larger automation scores in that study are hard to compare to our findings, since the underlying data is not
public.
40
Studies of earlier waves of technological change, such as the introduction of the internet and digital innovations the
workplace, demonstrate that such changes tend to put younger populations at an advantage, in comparison to older
workers.
31
productive benefits of the transition, for example through equipping workers with foundational
skills that can help them keep up with the changing character of jobs and drive the productive
character of such changes, rather than see their skills become obsolete. The fact that some
vulnerable groups, such as women and youth, face greater exposure to automation highlights the
importance of life-long learning so that that workers have the skills to adapt to changes in the
world of work. In the short-term, as shown in numerous ILO studies, social protection systems can
play an important role as macroeconomic shock stabilizers, and reduce the impact of transitions
for the affected workers and their households at the microeconomic level (ILO, 2023), especially
when their use is combined with skills development programs (ILO, 2023e). In the medium and
long-term, reducing gender gaps in the exposure to automation would require addressing factors
that perpetuate occupational segregation by gender, such as gender-based social norms (Carranza
et al., 2023).
At the same time, our findings show that the shares of jobs that could benefit from a productive
transformation with GenAI are consistently higher than those with automation risks across all LAC
countries, ranging between 8 and 12 percent of employment across countries. This is particularly
the case for the jobs in education, health and personal services. In addition, the sectors oriented
towards customer service (retail, trade, hotels, restaurants, etc.) face an elevated exposure to "the
big unknown", which means that a productive augmentation could also be sought in these jobs
with the right policies and incentives in place. Therefore, a tempting narrative that can be
constructed based on these statistics is that, in the big picture, more can be gained than lost as a
net job and economic effect of the transformation.
This is where our analysis provides new information to assess whether the lack of digital
infrastructure could be a buffer or a bottleneck to reap the economic benefits of GenAI. On the
one hand, our findings show that most workers exposed to GenAI automation are using digital
technologies, which suggests that the potential negative effects may not take long to materialize.
On the other hand, we find that inadequate digital infrastructure is a major bottleneck to realizing
the positive effects of augmentation, thereby impacting a significant segment of the labor force in
the LAC region. Nearly half of the occupations that could potentially benefit from augmentation
are hampered by digital shortcomings that will prevent them from realizing that potential.
Specifically, 6.24 percent of jobs held by women and 6.22 percent of those held by men are
affected due to these gaps. Similar limitations apply to the jobs in the “big unknown” category:
even though some of them could potentially pivot towards augmentation through increasing
complementarity between GenAI and the worker in these occupations, the digital gaps will prevent
large shares of these jobs from benefitting from such a scenario.
The extension of our finding is that the influence of digital divides would likely be even more stark
in regions of lower economic development than LAC, which underlines the need for equalizing
digital access in developing countries. Policies to achieve such goal should include not only those
related to digital infrastructure, but also those that aim to strengthen the incentives to adopt digital
technologies for a productive use (World Bank, 2016b) 41 to ensure that the transformative promise
41
World Bank (2016) provides several examples of policies not directly related to expanding the digital
infrastructure that promote the adoption of digital technologies by firms and people, such as fostering more
competition in product markets (both domestic and through international trade), improving the quality of the
educational systems, etc.
32
of GenAI does not bypass those who are most in need of its advantages. Focus on informality in
the transition will also be of key importance. We find that workers with formal jobs are more
exposed to GenAI automation than their counterparts in the informal sector. Even though formal
jobs typically offer some coverage of the social protection systems, eventual automation of these
occupations does not guarantee that the same workers would easily find new formal employment.
Since workers in the informal sector from developing countries rarely move to better paying and
formal jobs (Donovan et al., 2023), allowing automation to slowly erode the existing formal sector
would simply expand informality. Policies aimed at reducing the segmentation of the labor market
along the formal-informal lines would help improve the chances of displaced workers’
transitioning back to the formal sector. In this context, it is important to recognize that
technological transformation can actually offer opportunities for job e-formalization through
innovative government services (Chacaltana Janampa et al., 2024).
Finally, while this study provides a detailed overview of the LAC region, it is not without
limitations, some of which can be considered as open avenues for future inquiry. The first of those
concerns data on the use of computers and internet at work, for which the imputation from PIACC
was the best available strategy in our case. Obtaining new survey data for at least the most exposed
occupations would surely offer more precision to future estimates in that regard. In the absence of
such data, imputation based on the second cycle of OECD’s PIAAC, covering 2022-23, might be
a viable option, as soon as such data become publicly available. Second, more could be done to
understand how the task content of occupations varies across countries and in systematizing such
differences according to income groups and possibly regional characteristics beyond LAC. Third,
future studies could try to obtain more fine-grained characteristics of individuals’ internet access,
since network latency, reliability or type of devices used may affect the adoption and impacts of
GenAI by workers. 42 Accordingly, we will continue our efforts to collect data on firms’ and
workers’ adoption of GenAI in developing countries, in order to validate the exposure measures
developed in this study and to adapt our policy responses to more detailed findings. In that regard,
we would welcome collaboration with institutions, including national administrations, that might
be able to assist us in the collection of such data for future research.
42
Figure A5 shows a robustness check where exposure to GenAI is split by whether workers have access to internet
and a computer (instead of computer only).
33
References
Acemoglu, D., 2024. The Simple Macroeconomics of AI. Mass. Inst. Technol.
Acemoglu, D., Anderson, G., Beede, D., Buffington, C., Childress, E., Dinlersoz, E., Foster, L.,
Goldschlag, N., Haltiwanger, J., Kroff, Z., Restrepo, P., Zolas, N., 2023. Advanced
Technology Adoption: Selection or Causal Effects? AEA Pap. Proc. 113, 210–214.
https://doi.org/10.1257/pandp.20231037
Acemoglu, D., Anderson, G.W., Beede, D.N., Buffington, C., Childress, E.E., Dinlersoz, E.,
Foster, L.S., Goldschlag, N., Haltiwanger, J.C., Kroff, Z., Restrepo, P., Zolas, N., 2024.
Automation and the Workforce: A Firm-Level View from the 2019 Annual Business
Survey. NBER Chapters.
Acemoglu, D., Restrepo, P., 2022. Tasks, Automation, and the Rise in U.S. Wage Inequality.
Econometrica 90, 1973–2016. https://doi.org/10.3982/ECTA19815
Acemoglu, D., Restrepo, P., 2018. Artificial Intelligence, Automation and Work (No. w24196).
National Bureau of Economic Research. https://doi.org/10.3386/w24196
Adams-Prassl, J., Abraha, H., Kelly-Lyth, A., Silberman, M. ‘Six’, Rakshita, S., 2023.
Regulating algorithmic management: A blueprint. Eur. Labour Law J. 14, 124–151.
https://doi.org/10.1177/20319525231167299
Ananian, S., Aubert, P., Behaghel, L., 2006. Travailleurs âgés, nouvelles technologies et
changements organisationnels : un réexamen à partir de l’enquête « Reponse ». Suivi d’un
commentaire de Luc Behaghel : emploi des seniors - Des effets du changement
technologique aux recommandations. Econ. Stat. 397, 21–49.
https://doi.org/10.3406/estat.2006.7125
Aubert, P., Caroli, E., Roger, M., 2006. New technologies, organisation and age: firm-level
evidence*. Econ. J. 116, F73–F93. https://doi.org/10.1111/j.1468-0297.2006.01065.x
Autor, D., 2024. Applying AI to Rebuild Middle Class Jobs. Working Paper Series.
https://doi.org/10.3386/w32140
Autor, D., Dorn, D., Katz, L., Patterson, C., van Reenen, J., 2020. The Fall of the Labor Share
and the Rise of Superstar Firms*. Q. J. Econ. 135, 645–709.
Autor, D.H., 2015. Why Are There Still So Many Jobs? The History and Future of Workplace
Automation. J. Econ. Perspect. 29, 3–30. https://doi.org/10.1257/jep.29.3.3
Autor, D.H., Katz, L.F., Kearney, M.S., 2008. Trends in U.S. Wage Inequality: Revising the
Revisionists. Rev. Econ. Stat. 90, 300–323.
Azevedo, J.P.W.D., Inchauste Comboni, M.G., Sanfelice, V., 2013. Decomposing the recent
inequality decline in Latin America. World Bank.
Babina, T., Fedyk, A., He, A., Hodson, J., 2024. Artificial intelligence, firm growth, and product
innovation. J. Financ. Econ. 151.
Bakker, B.B., Ghazanchyan, M., Emmerling, J., Vibha, V., 2020. The Lack of Convergence of
Latin-America Compared with CESEE: Is Low Investment to Blame? [WWW
Document]. IMF. URL https://www.imf.org/en/Publications/WP/Issues/2020/06/19/The-
Lack-of-Convergence-of-Latin-America-Compared-with-CESEE-Is-Low-Investment-to-
Blame-49519 (accessed 3.26.24).
Brambilla, I., César, A., Falcone, G., Gasparini, L., Lombardo, C., 2023. Routinization and
Employment: Evidence for Latin America. Rev. Desarro. Soc. 131–176.
https://doi.org/10.13043/DYS.95.4
34
Brynjolfsson, E., Li, D., Raymond, L.R., 2023. Generative AI at Work. Working Paper Series.
https://doi.org/10.3386/w31161
Brynjolfsson, E., Rock, D., Syverson, C., 2021. The Productivity J-Curve: How Intangibles
Complement General Purpose Technologies. Am. Econ. J. Macroecon. 13, 333–372.
https://doi.org/10.1257/mac.20180386
Carranza, E., Das, S., Kotikula, A., 2023. Gender-Based Employment Segregation:
Understanding Causes And Policy Interventions.
Caunedo, J., Keller, E., Shin, Y., 2023. Technology and the Task Content of Jobs across the
Development Spectrum. World Bank Econ. Rev. 37, 479–493.
https://doi.org/10.1093/wber/lhad015
Cazzaniga, M., Jaumotte, F., Li, L., Melina, G., Augustus J., P., Carlo, P., Emma J., R., Mendes,
M., 2024. Gen-AI: Artificial Intelligence and the Future of Work [WWW Document].
IMF. URL https://www.imf.org/en/Publications/Staff-Discussion-
Notes/Issues/2024/01/14/Gen-AI-Artificial-Intelligence-and-the-Future-of-Work-542379
(accessed 2.19.24).
Chacaltana Janampa, J., de Mattos, F.B., García, J.M., 2024. New technologies, e-government
and informality | International Labour Organization. ILO Work. Pap. 112, ILO Working
Papers, Geneva.
Comunale, M., Manera, A., 2024. The Economic Impacts and the Regulation of AI: A Review of
the Academic Literature and Policy Actions [WWW Document]. IMF. URL
https://www.imf.org/en/Publications/WP/Issues/2024/03/22/The-Economic-Impacts-and-
the-Regulation-of-AI-A-Review-of-the-Academic-Literature-and-546645 (accessed
3.26.24).
Das, M., Hilgenstock, B., 2022. The exposure to routinization: Labor market implications for
developed and developing economies. Struct. Change Econ. Dyn. 60, 99–113.
Dieppe, A., 2021. Global Productivity: Trends, Drivers, and Policies.
Doellgast, V., O’Brady, S., Kim, J., Walters, D., 2023. AI in Contact Centers: Artificial
Intelligence and Algorithmic Management in Frontline Service Workplaces.
Donovan, K., Lu, W.J., Schoellman, T., 2023. Labor Market Dynamics and Development*. Q. J.
Econ. 138, 2287–2325. https://doi.org/10.1093/qje/qjad019
Egana-delSol, P., Bustelo, M., Ripani, L., Soler, N., Viollaz, M., 2022. Automation in Latin
America: Are Women at Higher Risk of Losing Their Jobs? Technol. Forecast. Soc.
Change 175, 121333. https://doi.org/10.1016/j.techfore.2021.121333
Eloundou, T., Manning, S., Mishkin, P., Rock, D., 2023. GPTs are GPTs: An Early Look at the
Labor Market Impact Potential of Large Language Models.
Erumban, A., Samaan, D., Van Ark, B., 2024. Latin America’s Productivity Puzzle: Insights into
diversification and structural change. Preseatation at the CEPAL conference:
https://www.cepal.org/sites/default/files/events/files/daniel_samaan_presentation.pdf.
Faverio, M., Tyson, A., 2022. What the data says about Americans’ views of artificial
intelligence. Pew Res. Cent. URL https://www.pewresearch.org/short-
reads/2023/11/21/what-the-data-says-about-americans-views-of-artificial-intelligence/
(accessed 3.14.24).
Felten, E., Raj, M., Seamans, R., 2021. Occupational, industry, and geographic exposure to
artificial intelligence: A novel dataset and its potential uses. Strateg. Manag. J. 42, 2195–
2217. https://doi.org/10.1002/smj.3286
35
Felten, E.W., Raj, M., Seamans, R., 2023a. Occupational Heterogeneity in Exposure to
Generative AI. https://doi.org/10.2139/ssrn.4414065
Felten, E.W., Raj, M., Seamans, R., 2023b. How will Language Modelers like ChatGPT Affect
Occupations and Industries? SSRN Electron. J. https://doi.org/10.2139/ssrn.4375268
Felten, E.W., Raj, M., Seamans, R., 2018. A Method to Link Advances in Artificial Intelligence
to Occupational Abilities. AEA Pap. Proc. 108, 54–57.
https://doi.org/10.1257/pandp.20181021
Frey, C.B., Osborne, M.A., 2017. The future of employment: How susceptible are jobs to
computerisation? Technol. Forecast. Soc. Change 114, 254–280.
https://doi.org/10.1016/j.techfore.2016.08.019
Garrote, S.D., Gomez, P.N., Ozden, C., Rijkers, B., Viollaz, M., Winkler, H., 2021. Who on
Earth Can Work from Home? World Bank Res. Obs.
Gmyrek, P., Berg, J., Bescond, D., 2023. Generative AI and Jobs: A global analysis of potential
effects on job quantity and quality (Working paper).
Goldman Sachs, 2023. Global Economics Analyst The Potentially Large Effects of Artificial
Intelligence on Economic Growth (BriggsKodnani), Global Economics Analyst.
Grampp, M., Brandes, D., Laude, D., 2023. Generative AI’s fast and furious entry into
Switzerland (Delloite).
Gutiérrez, G., Philippon, T., 2017. Declining Competition and Investment in the U.S. NBER
Work. Pap., NBER Working Papers.
Hui, X., Reshef, O., Zhou, L., 2023. The Short-Term Effects of Generative Artificial Intelligence
on Employment: Evidence from an Online Labor Market.
https://doi.org/10.2139/ssrn.4527336
ILO, 2023a. ILO Modelled Estimates (ILOEST database) [WWW Document]. ILOSTAT. URL
https://ilostat.ilo.org/resources/concepts-and-definitions/ilo-modelled-estimates/
(accessed 6.20.23).
ILO, 2023b. ISCO documentation: part III. Definitions of major groups, sub-major groups,
minor groups and unit groups. [WWW Document]. URL
https://www.ilo.org/public/english/bureau/stat/isco/docs/groupdefn08.pdf (accessed
6.20.23).
ILO, 2023c. World Employment and Social Outlook 2023: The value of essential work [WWW
Document]. URL https://www.ilo.org/digitalguides/en-gb/story/weso2023-key-workers
(accessed 7.31.23).
ILO, 2023d. World Social Protection Report 2020–22: Social protection at the crossroads – in
pursuit of a better future. ILO Flaghip Report.
ILO, 2023e. The ILO strategy on skills and lifelong learning 2030 | International Labour
Organization.
IMF, 2022. Productivity in Latin America and the Caribbean: Recent Trends and the COVID-19
Shock.
Lane, M., Williams, M., Broecke, S., 2023. The impact of AI on the workplace: Main findings
from the OECD AI surveys of employers and workers. OECD, Paris.
https://doi.org/10.1787/ea0a0fe1-en
Lewandowski, P., Park, A., Hardy, W., Du, Y., 2019. Technology, Skills, and Globalization:
Explaining International Differences in Routine and Nonroutine Work Using Survey
Data.
36
Lo Bello, S., Sanchez Puerta, M.L., Winkler, H.J., 2019. From Ghana to America : The Skill
Content of Jobs and Economic Development. Policy Res. Work. Pap. Ser., Policy
Research Working Paper Series.
Maloney, W.F., Molina, C.A., 2016. Are automation and trade polarizing developing country
labor markets, too ? Policy Res. Work. Pap. Ser., Policy Research Working Paper Series.
McKinsey, 2023. Economic potential of generative AI | McKinsey.
Messina, J., Pica, G., Oviedo, A., 2016. Job Polarization in Latin America. Inter-American
Development Bank, Washington, DC.
Messina, J., Silva, J., 2018. Wage Inequality in Latin America. World Bank Publ. - Books.
Noy, S., Zhang, W., 2023. Experimental evidence on the productivity effects of generative
artificial intelligence. Science 381, 187–192. https://doi.org/10.1126/science.adh2586
Nurski, L., Ruer, N., 2024. Exposure to generative AI in the European labour market:
distributional implications.
OECD, 2023. The impact of AI on the workplace: Main findings from the OECD AI surveys of
employers and workers (No. 288), OECD Artificial Intelligence Papers.
https://doi.org/10.1787/ea0a0fe1-en
OECD/PIAAC, 2019. Skills Matter: Additional Results from the Survey of Adult Skills.
Organisation for Economic Co-operation and Development, Paris.
Ozdeneron, Hakki, 2023. AI-Exposed Jobs Employ More Women. URL
https://www.reveliolabs.com/news/macro/ai-exposed-jobs-employ-more-women/
(accessed 6.26.23).
Peng, S., Kalliamvakou, E., Cihon, P., Demirer, M., 2023. The Impact of AI on Developer
Productivity: Evidence from GitHub Copilot. https://doi.org/10.48550/arXiv.2302.06590
Pew Research Center, 2023. Americans explain in their own words what makes them either more
concerned or more excited about the increased presence of AI in daily life. Internet Sci.
Tech. URL https://www.pewresearch.org/internet/2022/03/17/ai-and-human-
enhancement-americans-openness-is-tempered-by-a-range-of-concerns/ps_2022-03-
17_ai-he_01-02/ (accessed 3.14.24).
Prytkova, E., Petit, F., Deyu, L., Sugat, C., Tommaso, C., 2024. The Employment Impact of
Emerging Digital Technologies.
Restrepo, P., 2023. Automation: Theory, Evidence, and Outlook. Working Paper Series.
https://doi.org/10.3386/w31910
Rutgers, 2024. U.S. Workers Assess the Impacts of Artificial Intelligence on Jobs: Topline
Survey Results, Heldrich Center for Workforce Development.
Schady, N., Holla, A., Sabarwal, S., Silva, J., Yi, C., 2023. Collapse and Recovery: How the
COVID-19 Pandemic Eroded Human Capital and What to Do about It.
Svanberg, M., Li, W., Fleming, M., Goehring, B., Thompson, N., 2024. Beyond AI Exposure:
Which Tasks are Cost-Effective to Automate with Computer Vision?
https://doi.org/10.2139/ssrn.4700751
Webb, M., 2020. The Impact of Artificial Intelligence on the Labor Market.
WEF, 2023. The Future of Jobs Report 2023.
World Bank, 2016a. Poverty and Shared Prosperity 2016: Taking on Inequality, Poverty and
Shared Prosperity. The World Bank. https://doi.org/10.1596/978-1-4648-0958-3
World Bank, 2016b. World Development Report 2016: Digital Dividends (Text/HTML).
37
Appendix
Figure A 1. Comparison of TechXposure scores vs GBB scores (mean by occupation, z-scores)
Figure A 2.Comparison of Felten et al. (2023) ML scores vs GBB scores (z-scores)
38
Figure A 3. Labor market distribution in LAC countries by ISCO-08 2-digit occupations and sex
39
Figure A 4. Ranking of countries by the type of GenAI exposure
Figure A 5. Comparison of results on computer use between PIAAC (at work) and SEDLAC (at home) - augmentation category
Figure A 6. Jobs in augmentation category that do not use a computed at work: totals by country
40
Table A 1. Individual SEDLAC observations by country and year
Country Year Observations (Total) Observations (for AI Exposure)
Argentina 2021 98,822 37,640
Bolivia 2021 42,090 17,721
Brazil 2021 335,100 128,907
Chile 2022 202,231 75,936
Colombia 2019 756,063 316,870
Costa Rica 2019 34,863 13,690
Dominican Republic 2021 76,071 30,816
Ecuador 2021 30,026 13,480
Guatemala 2014 54,837 18,798
Honduras 2019 24,094 8,903
Mexico 2018 269,206 116,312
Nicaragua 2014 29,443 11,870
Panama 2018 39,218 16,395
Peru 2021 114,239 55,524
El Salvador 2021 64,937 26,036
Uruguay 2021 28,312 12,034
Total 2,199,552 900,932
Table A 2. Estimated coefficients of computer use at work from PIAAC
Dependent Variable
Computer at work Computer & Internet at work
isco_2d_c12 0.926*** 0.991***
isco_2d_c13 0.0273 0.0875
isco_2d_c14 -0.705*** -0.654***
isco_2d_c21 0.517** 0.383**
isco_2d_c22 -0.548*** -0.822***
isco_2d_c23 -0.686*** -0.686***
isco_2d_c24 1.690*** 1.669***
isco_2d_c25 3.406*** 2.387***
isco_2d_c26 -0.279 -0.292*
isco_2d_c31 -0.966*** -1.118***
isco_2d_c32 -0.873*** -1.155***
isco_2d_c33 0.674*** 0.614***
isco_2d_c34 -1.426*** -1.320***
isco_2d_c35 1.162*** 1.083***
isco_2d_c41 0.916*** 0.413**
isco_2d_c42 0.211 -0.247
isco_2d_c43 -0.419** -0.661***
isco_2d_c44 -0.819*** -1.076***
isco_2d_c51 -2.765*** -2.843***
isco_2d_c52 -1.804*** -2.059***
isco_2d_c53 -2.494*** -2.487***
isco_2d_c54 -1.903*** -2.143***
isco_2d_c61 -3.177*** -3.091***
isco_2d_c62 -3.201*** -3.169***
isco_2d_c63 -4.218*** -4.145***
isco_2d_c71 -3.570*** -3.525***
isco_2d_c72 -2.480*** -2.735***
isco_2d_c73 -2.011*** -2.246***
isco_2d_c74 -1.969*** -1.945***
isco_2d_c75 -2.967*** -3.098***
isco_2d_c81 -2.734*** -3.351***
isco_2d_c82 -2.796*** -3.458***
isco_2d_c83 -3.276*** -3.530***
isco_2d_c91 -4.347*** -4.434***
41
isco_2d_c92 -4.291*** -4.386***
isco_2d_c93 -3.234*** -3.657***
isco_2d_c94 -3.887*** -3.966***
isco_2d_c95 -3.527*** -3.669***
isco_2d_c96 -3.303*** -3.538***
Age 25-34 0.225*** 0.316***
Age 35-44 0.134*** 0.272***
Age 45-54 -0.162*** -0.00339
Age 55-64 -0.493*** -0.333***
Female -0.231*** -0.298***
High Education 0.893*** 0.948***
log_gdp 0.240*** 0.0915**
internet_users_rate 2.811*** 2.995***
broadband_rate 3.404*** 2.993***
Constant -3.377*** -2.172***
Observations 112,112 112,112
Note: SE not reported to preserve presentation space – available upon request.
42
Table A 3. Results of the pooled OLS with all individual observations, with country-level normalized population weights
Dependent Variable: AI Exposure
Augmentation Automation Big Unknown
Female -0.0313*** 0.0362*** 0.0770***
(0.00139) (0.000949) (0.00181)
Urban 2.81e-05 0.00615*** -0.00636***
(0.00165) (0.000717) (0.00185)
Age 25-34 -0.0105*** -0.00617*** 0.00144
(0.00187) (0.00122) (0.00231)
Age 35-44 -0.0104*** -0.0160*** 0.00173
(0.00189) (0.00122) (0.00235)
Age 45-54 -0.00920*** -0.0191*** -0.00332
(0.00204) (0.00123) (0.00244)
Age 55-64 -0.0129*** -0.0159*** -0.00331
(0.00227) (0.00130) (0.00271)
Medium Education 0.0110*** 0.0184*** 0.0406***
(0.00141) (0.000623) (0.00167)
High Education 0.0497*** 0.0203*** 0.129***
(0.00211) (0.00131) (0.00263)
Q2 income quintil 6.84e-05 -0.00187*** 0.0119***
(0.00157) (0.000717) (0.00187)
Q3 income quintil 0.00306* 0.00289*** 0.0251***
(0.00165) (0.000860) (0.00204)
Q4 income quintil 0.00826*** 0.00768*** 0.0410***
(0.00183) (0.000997) (0.00225)
Q5 income quintil 0.00768*** 0.00692*** 0.0640***
(0.00208) (0.00128) (0.00259)
Informal (legal definition) -0.0138*** -0.0187*** -0.0186***
(0.00149) (0.000874) (0.00172)
Employer -0.0223*** 0.00580*** 0.0205***
(0.00309) (0.00134) (0.00524)
Salaried Workers 0.0130*** 0.0314*** -0.0409***
(0.00216) (0.00107) (0.00348)
Self-employed 0.0168*** 0.00915*** -0.0325***
(0.00225) (0.00104) (0.00357)
Low tech. Industries 0.0459*** -0.0157*** 0.0392***
(0.00215) (0.00129) (0.00261)
Other industries 0.0846*** -0.00724*** 0.0405***
(0.00327) (0.00146) (0.00341)
Construction -0.00372** -0.00328*** 0.0222***
(0.00159) (0.000812) (0.00192)
Retail, wholesale trade, restaurants, hotels, etc. 0.111*** -0.00104 0.329***
(0.00187) (0.000840) (0.00257)
Elect., gas, water, transp., communication 0.147*** 0.0227*** 0.0721***
(0.00288) (0.00153) (0.00274)
Banks, finance, insurance, professional ss. 0.0693*** 0.0856*** 0.191***
(0.00260) (0.00250) (0.00353)
Public Adm., defense 0.0571*** 0.0339*** 0.145***
(0.00301) (0.00257) (0.00424)
Education, Health, personal services 0.295*** -0.0161*** 0.0155***
(0.00298) (0.00127) (0.00259)
Domestic ss. 0.0467*** -0.0422*** -0.0632***
(0.00217) (0.00116) (0.00230)
Constant 0.0193*** -0.0118*** 0.0297***
(0.00304) (0.00150) (0.00431)
Country and year fixed effects Yes Yes Yes
Mean of dependent variable 0.1203 0.0323 0.1821
Observations 888,685 888,685 888,685
R2 0.139 0.076 0.228
43