We often hear that artificial intelligence is important for economic growth, and while that claim makes intuitive sense, there isn’t a lot of hard data to back it up. A recent study from economists at MIT and Washington University in St. Louis offers some proof, though, showing how AI tools boost trade by allowing sellers to cross the language barrier.
Looking at data scraped from eBay, the researchers compared sales between the US and Spanish-speaking Latin American countries before and after the shopping platform introduced AI-powered translation for product listings in 2014. (Specifically, the translation tool affected the titles of listings and search queries, but not product descriptions.)
Prior to this eBay offered automatic translation, but the use of AI significantly improved the service’s accuracy. You would expect that better translations would lead to greater sales, and that’s exactly what the researchers found. Their data showed that sales from the US to countries affected increased 10.9 percent after the launch of the tool.
The researchers acknowledge that other variables could have caused this increase, but accounted for them in a number of ways. For example, they compared exports in this period to countries where languages not included in eBay’s translation tools are spoken, and controlled for factors like increased marketing spend and length of title listings.
The researchers also compared this data with eBay’s rollout of AI translation tools for Russian, Italian, and French. They found similar improvements, and estimate that the introduction of AI translation between countries is equivalent to reducing their distance from one another (a factor that has a big impact on trade volume) by 26 percent.
“These comparisons suggest that the trade hindering effect of language barriers is of first-order importance,” write the researchers. “Improved machine translation has made the eBay world significantly more connected.”
In a blog post from Washington University in St Louis, the researchers said their work could be used as a counterpoint pessimism about stagnating productivity rates around the world. MIT’s Erik Brynjolfsson, a leading expert in productivity who also co-authored this latest paper, has suggested that this could be caused by slow uptake of new technology.
“The problem is that it normally takes time for organizations to ramp up complementary innovations, be it organizational or technological, to harvest AI’s benefits,” Xiang Hui, a coauthor on the paper, told Washington University’s The Source. This eBay paper, says Hui, shows that when the tech is finally implemented, it can have significant effects.