The Taub Center is publishing a study that examines the level of exposure to artificial intelligence (AI) in the Israeli labor market and its expected effects. To this end, the researchers Michael Debowy, Prof. Gil Epstein, Prof. Benny Bental, Prof. Avi Weiss and Prof. Alex Weinreb used various indices to assess the expected impact of AI on the labor market in Israel.
The patterns of exposure found in the study make it possible to analyze and predict which occupations in Israel will be more affected by AI and which less and, accordingly, which populations are expected to gain and which are expected to lose.
Level of Exposure
To estimate the percentage of workers exposed to AI in a specific industry or population, various indices were used to assess the impact of the technology on specific occupations. The first index is based on the proportion of tasks out of all those that make up an occupation, which AI can purportedly perform. The score attributed to each occupation reflects its exposure relative to the average for all occupations in the economy. Combining this index according to the frequency of each occupation in different population groups (for example, industry or sector) provides an overview of the exposure level of workers in that population group relative to the average exposure of all workers in the economy.
Exposure by industry: Occupations in certain hi-tech sectors (finance and insurance, information and communication and professional, scientific, and technical services), as well as real estate, are much more exposed to AI than average, with over 90% of the workers in occupations with above-average exposure levels. In contrast, exposure to AI in construction, transportation, food services and agriculture is significantly below the average.
Exposure by education level: Unsurprisingly, the study found that as the level of education increases, so does the frequency of being in an occupation exposed to AI. It is particularly notable that workers with an academic education are much more exposed to AI than the average, while the exposure of workers who did not finish high school is much lower.
Additional cross-sections examined by the researchers include exposure by region (higher exposure in Tel Aviv, Petah Tikva and Ramat Gan), by workplace characteristics (especially high exposure among people who usually work from home), by age (highest exposure among workers aged 27–51) and over time (a significant increase in exposure between 2019 and 2021). An analysis by sector and gender after controlling for background variables found that men’s exposure is lower than women’s and that the exposure of workers in the Arab sector is the lowest.
Who will be harmed by AI and who will benefit?
Beyond the question of level of exposure to AI, it is important to examine the implications of this exposure: Which workers are likely to be harmed by AI and which are likely to benefit? To answer this question, the Taub Center researchers used a “complementarity index,” which measures the extent to which AI can serve as a complementary factor of production to human labor—thereby streamlining and improving work in that occupation and increasing worker productivity—or whether it puts workers at risk of replacement (i.e., it is a substitute factor of production).
The research shows that AI is expected to benefit workers in hi-tech, such as software developers. In contrast, an overwhelming majority of workers in the finance and insurance industry, which is also characterized by high exposure levels, are expected to be harmed. Thus, AI will replace many brokers and analysts in finance and insurance and even some accounting and legal professionals. In the education sector, most workers exposed to AI are expected to benefit, given that the development of relevant and updated tools will assist teachers in tasks such as preparing lesson plans and grading exams and assignments, while only a negligible percentage of teachers is expected to be harmed.
Educated workers exposed to AI are expected to gain, while less educated ones are expected to lose. Surprisingly, people who primarily work from home and are highly exposed to AI are equally divided between occupations expected to gain and those expected to lose. In contrast, a larger proportion of workers in highly exposed occupations who generally do not work from home is expected to benefit than to lose.
Although the exposure level is lower among Haredi (ultra-Orthodox) male workers and Arab female workers relative to non-Haredi Jewish workers and others, among those exposed to AI the ratio of workers in occupations exposed to AI who are expected to gain relative to those expected to lose is the highest. This ratio is also higher among Arab male workers and Haredi female workers than among the non-Haredi Jewish population.
Overall, it was found that about 30% of workers in Israel (about 1.3 million individuals) have a high positive exposure to AI, i.e. it is expected to improve and streamline their work, while the jobs of about 23% (approximately one million individuals) are expected to be threatened. Exposure is low for the remaining workers, and, therefore, the question of substitutability is not particularly relevant.
