Advances in artificial intelligence are rapidly transforming the world of work. This column investigates the effects of machine translation on (1) employment and wages in the translation sector, and (2) the demand for foreign language skills across various jobs and industries. Using variation in the use of machine translation across local labour markets in the US after the launch of the Google Translate mobile app, the authors find that areas with higher adoption of Google Translate experienced a decline in translator employment. The authors also show that improvements in machine translation have reduced the demand for foreign language skills in general.
Advances in artificial intelligence (AI) are rapidly transforming the world of work (Acemoglu et al. 2022, Brynjolfsson et al. 2025). As AI progresses, debates continue over whether this technology will complement human labour or displace it (Autor 2015, Frey 2019, Susskind 2020). In recent years, concerns have grown about AI’s impact on a range of professions. One striking example is the translation industry, where machine translation (MT) tools have seen rapid improvement. A 2024 survey found that over three-quarters of translators expect generative AI to adversely affect their future incomes, while others even question the enduring value of foreign language skills. Notably, The Economist recently remarked that “AI could make it less necessary to learn foreign languages”, a view echoed by OpenAI’s demonstration of Sky, which seamlessly translated speech between Italian and English in real time.
Yet, advancements in MT are not all that recent. The IBM701, the first MT system, was launched in 1954 in a collaboration between IBM and Georgetown University. Later milestones include free online translation services like Babel Fish in 1997 and Google Translate in 2006. However, it was the launch of Google Translate as an app on Android and iOS in 2010 – and its integration into browsers like Chrome – that triggered widespread adoption. As illustrated in Figure 1, searches for “Google Translate” spiked around this time, while queries for “Translator” declined correspondingly.
Figure 1 Google searches for “translator” and “Google Translate”, 2004-2024
Note: This graph plots monthly ‘interest’ in two Google search terms: “translator” (on the left axis) and “Google Translate” (right axis). The interest index is calculated by Google Trends using an unbiased sample of searches for different time periods and geographies.
In a recent paper (Frey and Llanos-Paredes 2025), we examine the impact of this shift on translator employment and the demand for foreign language skills.
Machine translation at work
We begin our analysis by constructing a city-level dataset using individual records from the 2010–2023 American Community Survey (ACS), which provides annual 1-in-100 samples of the US population. We then integrate these data with measures of the geographic spread of Google Translate based on search engine data, job postings data from Lightcast, and local employment and wage statistics for translators and interpreters from the Occupational Employment and Wage Statistics (OEWS) series.
Figure 2 Changes in searches for “Google Translate” across local labour markets, 2010-2023
Note: This map reports the 2010-2023 log change in internet searches for “Google Translate” across local labour markets.
In our study, we take advantage of the variation in MT uptake across 696 local labour markets in the US after 2010 (Figure 2). By comparing regions with high versus low adoption rates, we isolate the impact of MT on translator employment and wages. Recognising that changes in “Google Translate” searches might also be driven by other unobserved factors affecting translator employment, we address this potential endogeneity by instrumenting interest in Google Translate with local changes in “Google Drive” search activity. The rationale behind this instrument is that, although the two digital products serve distinct functions, both qualify as ‘experience goods’ (Chen et al. 2022). Recent research indicates that their adoption was primarily driven by the growth of Google’s brand awareness and reputation, rather than by other factors that could concurrently influence translator employment (Barwise and Watkins 2018).
Our findings indicate that regions with greater use of Google Translate experienced a growth slowdown in translator and interpreter jobs. In fact, for each 1 percentage point increase in MT usage, translator employment growth dropped by approximately 0.7 percentage points. Cumulatively, this effect translates into an estimated loss of about 28,000 new translator positions that might otherwise have been created over the 2010–2023 period.
Impacts on foreign language demand
The ripple effects of machine translation extend well beyond the translation industry. Traditionally, foreign language proficiency has been highly valued across sectors – from customer service and international business to healthcare and education. However, as MT accuracy and accessibility improve, this is changing. Our analysis of millions of job postings across local labour markets shows that regions with high MT adoption experience slower growth in job advertisements requiring foreign language skills.
This is true of all language pairs investigated: areas with robust Google Translate usage saw job postings demanding Spanish fluency grow by about 1.4 percentage points less than in other regions, with similar declines of roughly 1.3 and 0.8 percentage points for Chinese and German, respectively, and measurable dampening even for Japanese and French.
These effects are robust across various occupational categories, although the impact on Chinese language skills is relatively muted in IT, science, and engineering, suggesting that language remains important for technology transfer in these fields. Overall, our findings imply that as AI translation technology advances, the demand for bilingual skills is likely to continue its decline.
Broader implications: Translation and trade
These findings also carry potentially transformative implications for the future of globalisation. Historically, linguistic differences have posed significant challenges to trade, with research indicating that sharing a common language can boost bilateral trade by roughly 50% (Frankel and Rose 2002, Baldwin 2017). In many instances, the costs imposed by language barriers have been comparable to – or even exceeded – those of tariffs, quotas, and other formal trade restrictions.
In particular, improved machine translation could significantly boost global services trade, offering developing countries a new pathway for economic growth. Historically, many low-income countries have grown through industrialisation by capitalising on low labour costs to generate manufacturing exports. However, in recent decades, this manufacturing-led growth model has been increasingly challenged by automation and robotics, which, while boosting production, simultaneously diminish employment opportunities (Rodrik 2015). Consequently, tradeable services are emerging as a potential new driver of economic growth. Scholars such as Baldwin (2018) contend that future improvements in the economic fortunes of developing nations will be primarily derived from services trade rather than the traditional goods trade, a shift that is facilitated by advancements in digital platforms and AI translation technologies. By reducing language barriers, machine translation potentially enables billions of non-English speakers to participate in the global services marketplace, offering skills in engineering, design, marketing, consulting, and other areas.
Outlook
Whie the current effects of machine translation on translator employment and the demand for language skills have been moderate, they are likely to intensify as these technologies continue to advance. In particular, improvements in simultaneous speech interpretation pose a new frontier. In the past, interpreter work has been relatively insulated from automation; however, recent breakthroughs – exemplified by OpenAI’s demonstration of Sky – indicate that even real-time voice translation is beginning to encroach on this domain. These developments have significant implications for education policy, especially given that nearly 20% of students in American schools are enrolled in foreign language courses. As real-time voice translation becomes more refined, its labour market impacts, including potential effects on interpreters, warrant further investigation.
Source : VOXeu
