If you write something in Spanish, Julie will say you that “ is available in English, Spanish, German, and French. The same happens to Julie, the virtual assistant for the transport company Amtrak, who is able to help customers book reservations, navigate and plan a vacation. This is possible, unfortunately, just in English. This fact is keeping the Hispanic population away from the bank. However, if you write something in Spanish she will return a nice “Sorry, I can’t speak your language yet, but my colleagues are working on it”.
When you talk to Erica, the artificial intelligence chatbot of Bank of America, she can understand most of the conversation in English. Not even such large companies as Bank of America or Amtrak, with plenty of resources, have chatbots speaking in Spanish. This language issue is hampering companies when developing a multilingual automated customer support channel that meets their customer speaking in different languages. That’s why, probabilistic models are in need of a linguistic approach to understand different languages, also if they are as complicated as Spanish, Finnish, Arabic or even Hindi. Apart from the verb issue, there are much more difficulties that come with Spanish, that slow down a Machine Learning process. In contrast, the equivalent verb in Spanish presents a wide range of variations. Think about the verb to give, that is just composed by the following 5 forms give, gives, gave, given and giving. English, for example, has very few different forms for each word.
Every language is different and this training process can vary, in terms of difficulty, from one to another. Making bots understand users in any language is not an easy task. What is the problem with understanding Spanish? Spanish is the second most spoken language of the United States by far and is the third-most used language on the internet, after English and Chinese. In the North American market, for instance, where a 20% of the population speak in Spanish, there is a big need of chatbots supporting that language not to lose any potential client. One of the biggest is, in this context, the ability to understand different languages. Nevertheless, some obstacles are standing in their way. However, they are being stopped by some hurdles in the way. Is it that difficult to make AI understand Spanish?Ĭompanies are investing large amounts of money in getting a so-called chatbot automation for their customer support channels. However, large companies like Bank of America or Amtrak are struggling to have a Spanish-speaking customer support bot to multiply their sales opportunities. Up to 60 million people speak Spanish today in the USA.
0 kommentar(er)
