“What you don’t understand, the machine understands. You know, machines know better. This is precisely the manifestation of artificial intelligence surpassing humans in many fields, and it is also the reason why it is necessary for us to give many jobs to artificial intelligence.

“

What you don’t understand, the machine understands. You know, machines know better. This is precisely the manifestation of artificial intelligence surpassing humans in many fields, and it is also the reason why it is necessary for us to give many jobs to artificial intelligence.

Language can be one of the biggest barriers in international trade. As an important application scenario of artificial intelligence, machine translation is moving towards solving this problem.

Recently, researchers from the National Bureau of Economic Research, the Massachusetts Institute of Technology and the University of Washington released a paper analyzing how the e-commerce platform eBay has achieved the growth of international trade through machine translation.

Research shows that with the help of machine translation systems, international trade exports on the eBay platform increased by 17.5%. According to the authors of the paper, this analysis shows “a direct causal link between AI applications and economic activity.”

At present, transactions on the eBay platform cover more than 200 countries, 20% of transactions are cross-border transactions, and one in every three new users is completed through cross-border transactions.

Obviously, when customers from all over the world search for massive product information on the eBay website, the difficulty of translation work can be imagined. Breaking down the communication barriers of multinational languages ​​and providing buyers and sellers with a real-time and seamless native language experience is exactly why machine translation was introduced on the eBay platform.

Earlier, Juan Rowda, eBay’s machine translation language expert, said in a 2016 article that “human translation is not feasible for companies.” Because, there are about 800 million listings on the eBay platform, and each listing contains about 300 words. “Even if only 60 million Russian-language listings were translated, it would take 1,000 translators five years,” Rowda said.

According to reports, eBay introduced a machine translation (eMT) engine in 2014, which was first opened to users in Russia, South America, and the European Union.

On the eBay platform, machine translation engines are not used for product categories, advertisements, or other fixed areas, but specifically for search.

Buyers can search in Spanish, and the translation tool can match search queries to listing titles stored in English and Display search results directly in Spanish, so Spanish-speaking users don’t need to know English when they land on the site to search for products.

The paper’s data show that “the introduction of machine translation systems significantly increased international trade on this platform, with exports increasing by 17.5%.” Many different control samples were also used in the study, but the data ranged from 11.8% to 20.9%. % varies. In other words, the trade growth brought about by machine translation is unquestionable.

The study also found that machine translation systems had a greater impact on products that were differentiated, less expensive, or contained more words in product listing titles, and were especially effective for buyers with less online shopping experience.

Another interesting finding is that consumers appear to “benefit more from machine translation than sellers…because consumers benefit both from the reduction in language friction and from lower prices.”

After introducing its own machine translation tool, eBay has reduced translation-related search costs and improved search match accuracy and listing title quality due to improved translation quality.

Much of the changes came from eBay’s June 2014 acquisition of machine translation technology company AppTek. The acquisition provides eBay with relevant talent in machine translation technology and speech recognition.

With the continuous advancement of artificial intelligence, including machine learning, machine vision, speech technology, and machine translation technology, it may be more convenient for us to shop in the future.

At that time, we only need to use our voice to tell our AI assistant what we need, and he (or she) will be able to give you the best choice based on your preferences, habits, shopping experience and other data.