Wikipedia’s goal is to make the world’s knowledge accessible to everyone on the planet, but it would be the first to admit that its efforts are skewed somewhat toward those who read English. To help fix this, the Wikimedia Foundation (which runs Wikipedia) announced today that it’s partnering with Google to take advantage of the company’s AI translation skills.
Google Translate will be integrated into Wikipedia’s in-house translation tool at no cost, added as an option alongside the open-source translator Apertium, which has been used to translate some 400,000 Wikipedia articles to date. In both cases, the software does the first pass of translating an article before a human editor steps in to correct any mistakes.
Wikipedia editors have been asking for access to extra translation tools like Google’s for a while. The ad company’s software is widely acknowledged as some of the world’s most advanced, thanks in no small part to Google’s AI prowess. And — as a more obvious benefit — it can handle 15 more languages than Apertium can.
These extra languages include Zulu, Hausa, Kurdish (Kurmanji), and Yoruba. These may not be the best-known languages on the world stage, but obviously, that doesn’t at all diminish their importance to those who speak them. Zulu, for example, is spoken by some 12 million people, but there are only around 1,000 Wikipedia articles available in the language.
In an FAQ, the Wikimedia Foundation has answered worries that some editors may have about the partnership, noting that no personal data is shared with Google, the translated content will still be freely available under a creative commons license (like all Wikipedia articles), and there will be no Google branding added to the site as a result of the deal. The agreement is only for a year, and it will be reevaluated after that, and the Wikimedia Foundation can terminate it at any time.
Importantly, all translations generated in this way will be freely available to the public. That means they can be used to improve other translation tools, including open-source software like Apertium. After all, if you want to make the world’s knowledge available freely to all, it’s a good idea to empower people to spread that knowledge themselves.