量化人工智能在拉丁美洲和加勒比地区的就业潜力

JOBS OUTCOMES STORY RESULTS Analysis found that30–40% of jobs in Latin Americaand the Caribbean are exposed in some way to gen-erative artificial intelligence(Gen AI), with exposurehighly linked to a country’s economic status. Exposedjobs are more likely to be urban, formal, and high-er-paying, and to require higher education. Quantifying theJobs Potential ofAI in Latin Americaand the CaribbeanPublic Disclosure Authorized Between 8 and 12% of jobs in the region could see aboost in productivityby harnessing Gen AI – but upto half, some 17 million jobs, won’t be able to lever-age the potential benefits because of a lack of digitalinfrastructure. AT A GLANCE Gen AI puts2–5% of jobs in the region at risk ofautomation,disproportionately impacting younger,educated, urban workers, and especially women. Onaverage, women workers are twice as likely to be atrisk of automation from Gen AI. REGION/COUNTRYLatin Americaand the Caribbean PROGRAM/PROJECTPublication:Buffer or Bottleneck? EmploymentExposure to Generative AI and theDigital Divide in Latin AmericaPublic Disclosure Authorized Between13% and 22% of workers are exposed to GenAI in contexts that could lead either to automationor augmentation,depending on the evolution of thistechnology, workers’ characteristics, and comple-mentary policies. THE CHALLENGE Generative artificial intelligence (Gen AI) is set to potentially transform the labor market, presenting both op-portunities and challenges. Studies about its possible impacts have generally focused on high-income coun-tries. But AI might be a crucial path to greater productivity in emerging markets – or, like other recent wavesof technological change, it might widen the gap between low- and high-income workers. These effects are ofparticular concern to Latin America and the Caribbean (LAC), one of the world’s most unequal regions, andone that has struggled with a persistent productivity gap, in large part because of barriers to innovation andtechnology adoption. To effectively minimize the risks and leverage the benefits of Gen AI, LAC countries wouldneed to understand their economy’s occupational exposure to the technology – yet that type of assessmenthad never before been undertaken in the region. WBG APPROACH In partnership with the International Labor Organization, the World Bank produced the first analysis of the LatinAmerican labor market’s exposure to Gen AI and what it means for the hundreds of millions of workers acrossthe region. The policy paper leveraged a rich set of harmonized household and labor force surveys, breaking down– by country, by demographic, and by sector – the jobs at risk of automation, those where AI could boost pro-ductivity, and those that might fall in between. One major consideration for this kind of analysis is that develop-ing economies tend to adopt technology more slowly, a variable that the report conscientiously accounted for.Critically, all the data on the country level has been made publicly available online. This data is foundational tocountry-specific insights, informing policy responses that can leverage AI to benefit workers, grow economies,and increase prosperity for all. LESSONS LEARNED One of the analysis’s most urgent lessons is the need to equalize digital access in developing countries. Inade-quate digital infrastructure prevents countries – especially lower-income countries – from capitalizing on GenAI, making it a major bottleneck for increasing productivity. Additionally, it is crucial to scale up the collectionof high-frequency labor market data in developing countries, to move from measuring Gen AI exposure to mea-suring its impacts. Such data would allow the early detection of labor-displacement effects and inform evi-dence-based labor market policies. NEXT STEPS A key step will be to support governments in creating policies that minimize Gen AI’s disruptions and maximizeits benefits: strengthening job protection measures and social protection systems, equipping workers with skills,expanding digital infrastructure, and reducing the digital divide. Future work could also dig into the details anddata gaps that limited this analysis. This includes collecting better data on the use of technology at work, on thecharacteristics of workers’ internet access, or on how the tasks of certain types of jobs vary across countries.These methods will be updated and scaled up globally for the next World Development Report on AI.