Humans Defeat AI in Translation!
Matt Nelson
? Owner | Chief Strategist I Help Business Owners Outrank and Optimize through Compelling Content ??????????
On Tuesday 21 February, the International Interpretation Translation Association hosted a competition between four professional human translators and three language translation programs powered by Artificial Intelligence (AI). The programs were provided by Google Inc., Naver Inc. (South Korea's top internet provider), and Systran International, a leading automated interpretation company.
Translation language pairs for both sides of the competition included translation of articles from English to Korean and Korean to English. Articles consisted of both fiction and non-fiction. The AI and human competitors were provided 50 minutes to translate the content, which was then evaluated by two professional translators.
The Results...
In an astounding victory, the humans beat the AI team in the Korean to English translation category with an average score of 25 out of 30. The AI team's average score was between 10 and 15 out of 30.
Last year, both Google and Naver rolled out services using neural machine translation (NMT), an AI translation system which learns from millions of examples across more than 100 languages.
The Future of AI Translation...
The problem with the AI team's performance was in contextual understanding of the articles. Organizers stated that over 90% of the AI translations were "grammatically awkward." According Systran, this gap in technology will soon be filled since the new AI translation technology can learn without being explicitly programmed...
In a forum held by the association, Kim Yoo-seok, director of Systran, stated "I would say the technology is at the stage of elementary school at the moment but it will improve to the level of high school and college just in one or two years."
In a world of ever-blurring linguistic boundaries and a more globally-diverse economic playing field, the role of the translation industry is more important than ever.
With the rise of big data and analytics, machine translation has taken over some of the painstaking burden from human translators. In the meantime, professional human translation still has a huge market share, particularly in the fields of medicine, science and technology.
The real question is, will artificial intelligence be able to replace the contextual art of human translation?
What are your thoughts? Comment and share :)
Please read the full article from Yonhap News here.
Front-End Developer at Freelance Web Designer (Self-employed)
7 年Good article:)
人工智能技术总裁
7 年cool article, I see AI translation more as a tool for human translation. Most human translators have already been using some form of machine translation - it's more efficient, takes less time, etc. AI just makes that machine translation better. Humans can focus more on the context, the professional message, the human feel that we all connect with when we communicate.
Project Manager at Type 2 Translate LLC
7 年I highly doubt machine and/or AI translation could achieve the complexity of the human translation skill within one, or even a few years. I would even venture to guess, it would take decades to reach the skills of a talented human translator - but I may be bias ;)