A pair of teams from Microsoft and Alibaba made the headlines in tech news this month after creating AI programs that can outperform humans in a reading comprehension test, but what exactly does reading mean?
“Think about it like this: for each question, the computers and humans know that the answer has to be in the source paragraph somewhere — and not just the answer, but the exact wording. Asking “Whose authority did Luther’s theology oppose?” seems tough, but when the source text includes the sentence “[Luther’s] theology challenged the authority and office of the Pope,” it doesn’t look quite so bad. You don’t need to understand what “authority” is, you just need to look for basic grammatical components, like the subject and object of a sentence.”
These methods, although proven to be successful, can easily be tricked by adding adversarial examples (by making the system misclassify data). It seems that computers aren’t quite there yet.
Source: THE VERGE
“The gap between the rich diversity of language spoken in the offline world and languages in cyberspace remains.”
Internationalized domain names that use non-Latin alphabets, including Cyrillic or Greek, have been declining. Multilingualism advocates urge software companies to recognize these alphabets in order to encourage more content in minority languages.
“ Internet platforms have been especially helpful in getting people to use less widely spoken languages online.”
Japan’s Internal Affairs and Communications Ministry plans to improve automatic translation system using AI by better defining the context of the situation.
“The envisaged AI technology will be capable of scanning environments to collect more information, such as facial expressions, the date and time, and the place a conversation is taking place, making it possible to generate more accurate translations.”
Source: THE STRAITS TIMES
This article is part of our Weekly Roundup, where we provide you with the latest news on Localization, Translation, and Language. For questions or comments, please click here.